PROCEEDINGS PAPER

Design and Constructionof Hudson Hope Bridge

by J. Dudra*

INTRODUCTION

The Hudson Hope Bridge, whichcrosses the Peace River in northernBritish Columbia, is located betweenthe town of Hudson Hope and thegiant Peace River Power project.The site is approximately 600 milesnorth of the Canada-United Statesborder and is located in the PeaceRiver Valley, east of the RockyMountains. The crossing is on thenewly constructed Chetwynd-Hud-son Hope highway which providesa shorter link for traffic movingfrom central British Columbia to theHydro Electric project. Eventually,when additional highway construc-tion is completed, the above willprovide a shorter route connectingto the Alaska Highway.

In 1961, the British Columbia De-partment of Highways appointedPhillips, Barratt and Partners, con-sulting engineers, to conduct sitesurveys for a feasible crossing of thePeace River upstream from HudsonHope, and to proceed rapidly withthe final design of a bridge struc-ture which would facilitate the pas-sage of heavy equipment and mate-rial to the Peace River HydroProject.

It was at this time that Col. H. H.Minshall, a well-known bridge erec-tor in British Columbia, conceived

°Partner—Phillips, Barratt and Partners,Consulting Engineers, Vancouver, B.C.,Canada.

the idea of a precast concrete sus-pension bridge. This concept ap-peared to be particularly suitable tothe Peace River site which was richin local concrete aggregates, far re-moved from railhead and deprivedof good highway access. Conse-quently, the consultants were re-quested to give consideration to theMinshall concept if the designproved workable and if the con-struction appeared economically fea-sible.

The initial surveys disclosed sev-eral potentially feasible crossingsites, the most economically favor-able of which was in a constrictedarea of the river where both bankswere nearly vertical. Design studiesof a precast concrete suspensionbridge to span the river at this siteconfirmed both workability andeconomy and this concept was there-fore adopted for the final design ofthe crossing.

Eleven tenders were received forthe construction of this bridge andthe contract was awarded to the lowbidder, Hans Mordhorst Ltd., inSeptember 1963. The structure wascompleted and opened to traffic oneyear later in September 1964 (Fig.1).

SITE AND SUBSURFACE CONDITIONS

The site is in a portion of the riverwhich is considered part of the Low-er Peace River Canyon. The riveroccupies a rock walled valley which

52 PCI Journal

Fig. 1—Completed Bridge

has several constrictions betweenthe dam site and a point about 3.3miles upstream of Hudson Hope.The chosen site is the second con-striction upstream of Hudson Hope.

Close to the river, the overburdenis fairly shallow. The underlyingstrata consists of varying grades ofsedimentary rock layers, which arecomposed of sandstones and sandyshales at the surface, and shales inthe deeper portions.

These sedimentary rocks are mem-bers of the Fort St. John group ofthe Lower Cretaceous Age and as aresult have been "prestressed" andfairly well cemented. However, theydo exhibit a development of shalyweathering when exposed to air. Atsome locations, where the river flowimpinges on the banks, erosion or

scour, chiefly of the shale, takesplace. In colder weather, frost ac-tion loosens the outside layers ofrock. This action, quite commonlyproduces a 3 to 5 ft. projection oroverhang of the uppermost layersof rock. For the above reasons, anyfoundations above the canyon wallshad to be located more than 40 ft.away from the upper edge of thewalls.

The Peace River, one of the ma-jor rivers in British Columbia, ispresently being prepared for thePortage Mountain Dam, some 8miles upstream of the crossing. Untilcompletion of the dam, when dis-charge will be stabilized and keptfairly uniform, the river dischargevaries considerably between theminimum and maximum. At the

April 1966 53

crossing site the low water elevationof 1503 ft. provides a channel depthof 40 ft.; whereas, at the high waterelevation of 1527 ft., the channeldepth becomes 64 ft. Winter freeze-up conditions have been known toproduce an 18 ft. thick ice layer onthe surface in some locations. Springbreakup of the ice causes a severebuild up in the constricted channelportions of the river.

ALTERNATIVE CROSSINGSSite surveys indicated two possi-

ble locations for a crossing over theriver. Site A, the first constrictionupstream of Hudson Hope, had adistance of 670 ft. face to face ofbanks. Site B, 3/4 of a mile furtherupstream had a distance 565 ft.between the near vertical banks. TheChetwynd-Hudson Hope highwaydesire line passed east of both cross-ings. Therefore, Site B, when eval-uated against Site A, had to becharged with an additional one anda half miles- of highway. Four alter-native crossings were considered:

Scheme 1—Fixed steel deck archat Site B, 565' c/c ofskewbacks

Scheme 2—Continuous steel decktruss at Site B, 160'-340'-160' main spans

Scheme 3—Concrete box girdersuspension span at SiteB, 675' c/c of towers

Scheme 4—Continuous steel decktruss at Site A, 150'-340'-270' main spans

Scheme 3 was a modified schemeof a segmental concrete box girdersuspension bridge conceived by Col.H. H. Minshall.

Schemes 2 and 4 had their mainpiers located within the channel con-strictions.

The fixed steel deck arch and theconcrete deck suspension bridge

were not actively considered at SiteA because of less favorable riverbank elevations and significantlylonger span requirements.

In summary, the results of theeconomic investigations revealed thefollowing. Costs shown were onlyrelative, and did not include bridgedetails, approaches and other por-tions common to all schemes.

Site BScheme 1 (Steel arch) $1,321,000Scheme 2 (Steel truss) $1,357,000Scheme 3 (Concrete

suspension) $1,275,000Site A

Scheme 4 (Steel truss lesscredit of $206,000 forshorter highway) $1,343,000

From the foregoing cost data, thesuspension scheme offered a slightsaving over the other schemes con-side red. Additional considerationswere as follows:

1. The desirability of keepingpiers and thrust blocks out of a riverwhich is being developed for hydroelectric power so close upstream.

2. The problem of potential longterm scour around piers or thrustblocks which are located in a nar-row river constriction.

3. The indeterminacy of ice build-up and ice loads on river piers in thenarrow channel.

4. The relatively short construc-tion period during which river pieror arch thrust block work could safe-ly be carried out in this area. It maybe noted that the river peak flowoccurs during the period when idealconstruction time prevails.

On the basis of the above con-clusions, Scheme 3 was adopted asthe choice for the crossing. Thechoice was a sound one, as the con-crete box girder suspension bridgeutilized native materials to a verylarge extent. The area containedgood aggregates for the concrete

54 PCI Journal

deck, towers, and anchor housings.The sedimentary rock strata on eachbank provided excellent support forthe tower footings and relatively lowcost anchorages for the main cables.

The successful completion of thisproject at a cost slightly below theconsultant's estimates, served to con-firm the preliminary studies on thisproject. Fig. 2 shows the elevationof the crossings.

BASIS OF DESIGNThe requirements for the deck

and its supports as stipulated by theDepartment of Highways were asfollows:

1. Roadway width—curb to curb28'-0" for two 14'-0" lanes.

2. No sidewalks—curb to handrailface-10 in.

3. Design live load:a. H25-S20 for slab, cross gir-

ders, suspenders and im-mediate connections.

b. H20-S16 for stiffening gir-der and main cable.

4. Design Specifications—"Specifi-cations for Highway Bridges"by British Columbia Depart-ment of Highways and "Stand-ard Specifications for HighwayBridges"-1961 edition by theAmerican Association of StateHighway Officials.

ANALYSIS OF STIFFENING GIRDERThe deck cross-section was given

considerable study. The heavy wheelloads (with possible overloads ofconstruction equipment) limited thepractical span of the slab if the deadload of the deck was to be keptminimal. After several trial sectionswere evaluated, together with theiradded function as a longitudinal stif-fening element, the section as shownin Fig. 3 was chosen as the one bestfulfilling all the requirements.

The analysis of the longitudinalstiffening girder was based on theDeflection. Theory as developed byDr. D. B. Steinman.' For prelimi-nary design of the structural compo-nents an A.S.C.E. paper by Hardestyand Wessman2 was utilized.

The decision to use the deflectionrather than the elastic theory forthe analysis of this structure wasmade primarily because the formertakes into account the effect of thedeadload of the suspended deckcomponents in partially overcomingthe bending moments due to liveload. Admittedly, the elastic theory,based on superposition of combinedloading on the structure and theutilization of influence lines involvesa simpler analytical process; and,where the deck dead load is a rela-tively small increment of the total

a

' 8a' Olear

192- 0" 680- O" 2O7'- 0"

Fig, 2—Elevation

April 1966 55

GALV. BRIDGE ROPESUSPENDERS(4 PER DECK UNIT

37-0'

28'-O" ROADWAY

6' CONCRETE SLABPLUS 2 4 TOPPING

SLOTS AND 5'" CONCRETE SLABROLLERS 8" CONCRETE WEBSFOR POSTTENSIONING 29-6"STRANDS

Fig. 3—Typical Section—Suspended Precast Deck Unit

load, this approximate theory is un-derstandably adequate. However, inthe case of a suspended concretestructure, the dead weight is a sig-nificant load factor which, if ignoredin the live load analysis, wouldmake the design of a concrete sus-pension structure completely im-practical.

A further differentiation from con-ventional steel suspension bridge de-sign should also be noted.

In conventional suspensionbridges, the stiffening element hasbeen the truss or girder, which islocated directly below each cable.The trusses span longitudinally andare supported elastically by the sus-penders. The floor system only dis-tributes the load outward to thetrusses and is ineffective as a longi-tudinal stiffening element.

The box girder, by its two-wayload distributing action of the longi-tudinal webs, flanges and cross-girders restrains the loads in a moreefficient and safer manner. Because

of this, the concrete box girder deckcan economically and functionallycompete against the conventionalsteel stiffening truss for spans below800 feet and possibly longer if fur-ther improvements are made.

Fig. 4 shows the geometries ofthe behavior and functions of astiffening girder under the action ofa particular live load. Observing thefigure, and remembering the law ofvirtual work, it is evident as to thepart the dead load performs in con-tributing to the restraint of the liveload moments.

In general, the properties of thestiffening girder were arrived atfrom the following requirements:

1. Transverse bending set the topslab thickness of 61/z in.

2. Bottom slab thickness of 5½ in.was set by the Code.

3. Web thickness of 8 in. was theminimum that was felt to be prac-tical for "on site" construction.

4. The depth, d, was a variablewhich required several trial analyses

56 PCI Journal

before it could be established. Alarge d produced high unit stressesin the flanges. A small d produceda very flexible stiffening girder whichapproached the unstiffened cablecondition. In addition, a small dprevented the creation of cell open-ings in the box girder through whichconstruction and maintenance per-sonnel could pass. Various trial de-signs finally confirmed that the mostfunctional depth for this span was4'-O" combined with a cable sag of90 ft. (It is interesting to note thatthe economic and functional ratio ofcable sag to span, n, for steel sus-pension structures is between 0.09and 0.11, whereas the feasible ratiofor this structure was 0.13.)

Fig. 5 shows the live load and itsposition, which produced the maxi-mum moments along the girder.

The behavior and analysis of theconcrete box section as a girder toresist the live load bending momentswas directly related to the stiffnessof the girder. This stiffness is notonly a function of the moment ofinertia, I, of the girder but also ofthe modulus of elasticity, E, of theconcrete. In conventional steel sus-pension bridges, the E of steel re-

gardless of grade is fairly definitiveand constant. However, this is notthe case with concrete, whose vari-able modulus of elasticity is affectedby creep and flow and whose de-flections therefore, are dependenton whether the loading is short orlong term.

A study was therefore necessaryto establish the relationship betweenthe variations in the concrete mod-ulus and the bending moments inthe concrete stiffening girder. Thisrelationship is shown in Fig. 6. Anexamination of the results of thisstudy revealed that the variations inthe girder moment were not directlyproportional to E but still greatenough to introduce significant er-rors into the analysis of the girder.Obviously, overstress could only oc-cur if the in-situ modulus E, wasgreater than the design modulus. Asa safeguard against the possibleoverstress, a layer of reinforcedneoprene was placed between eachsegmental concrete deck unit. Thismaterial, with its very low modulusof elasticity, served to achieve thefollowing:

1. Reduced the modulus of elas-ticity of the composite girder

^\ INITIAL CABLE POSITION //

•.\

UNSTIFFENED CABLE //

FINAL CABLE POSITION ///

v T\^^^^___^^^/ - \ INITIAL GIRDER POSITION

v POSITION of FLEXIBLE GIRDER

Live Load v FINAL GIRDER POSITION

kl 4

Fig. 4—Deflection of cable and stiffening girder under partial live load

April 1966 57

FOR ALL LOAD CONDITIONS P= 39.6k

WI = 1.41,1ftWd= 9.9 k/ft.

1 WI

270'TEMP.+110° F

M.= 7800-k-ft

Fc =+1000 psi.

1080 psi.

MAX. VE B.M. in GIRDER

WI

390'TEMP.-50°F

M. t.= 6200 k-f

Fc =-800 psi.

+860 psi.

MAX. -VE B. M. in GIRDER

WI

180TEMP.+110° F

M. = 6300 k -fl.Fc=+810 psi.

-870 psi.

MAX. + VE 8. M..t Ck

WI WI

IIIIIIIIJ

240' 240TEMP.-50°F

M. =4800 k-ft.

Fc=-620 psi.

+660 psi.

MAX. -VE B. M. at Lk

Fig. 5—Live Load Positions for Maximum Moments

mrri

Q Z

F- NOwQ D

.25 .50 .75 1.0 1.25 1.50 1.75 2.00

ACTUAL 'E'DESIGN 'E'

Fig. 6—Relationship between Girder Modulus and Bending Moment58 PCI Journal

as a safeguard for short termlive load conditions.

2. Functioned as a "spring" orplastic hinge in the event ofheavy load concentrations onthe bridge.

3. Protected the abutting faces ofthe deck units during erection.

The design thickness of the rein-forced neoprene, V4 in., was estab-lished by analysis of the requiredcomposite modulus which, conserva-tively, was assumed to be a com-bination of the E for the neopreneand a value of the concrete modulus50% greater than normal.

The stiffening girder was dividedinto 34-20 ft. long segmental units.Each unit weighed 90 tons. A crossgirder located at each end of theunit served as the connection tothe suspenders of the main cablesystem. By this system, as shown inFigs. 3 and 8, each segmental unitwas supported by 4 suspenders. The

units were interlocked not only bythe longitudinal prestress force, butalso by 6 curved keys, one in eachweb. The latter provision also facili-tated proper alignment of the unitsduring and after erection.

An operation with 34 precast seg-mental units could result in neardisaster if the joints were notmatched exactly. The use of groutedjoints on a suspension span which ex-periences considerable vertical andlateral movement was consideredunsatisfactory. The solution there-fore was to cast the units on a bedset to a predetermined profile andwith their respective faces abutting.Detailing prevented the adjacentcross girders from bearing on eachother. The reinforced neoprene, ce-mented to one abutting face, servedas a form for its mate. Sixteen unitswere cast on each bank (Figs. 7and 8) and after erection, the twoend units, which contained the end

Fig. 7—Segmental Units in Casting Bed

April 1966 59

Fig, 8—Segmental Units in Casting Bed

blocks and spiral anchorages werecast in place.

After erection and post-tensioningwas completed, the deck slab wasprotected by a 2 in. thick layer ofepoxy bonded concrete topping. Atransverse joint was placed over thegirder joint to prevent structural ac-tion of the topping with the stiffen-ing girder.

AERODYNAMIC STABILITYThe areodynamic stability analysis

for this bridge was based on anA.S.C.E. Proceedings paper by D. B.Steinman.3

Findings indicated that the struc-ture was stable for both vertical andtorsional oscillations—firstly, becauseof its relatively high dead load andsecondly, due to the shape of thedeck. The coefficient of rigidity, thestiffness ratio and the aerodynamicstability constant, all had valueswhich exceeded the suggested ones

by an adequate margin. Diagonalsbetween the stiffening girder andthe main cables at the center of thespan prevent relative longitudinalmovement and thus help to inhibittorsional motion. Site observationssince completion of the bridge haveconfirmed that the vibrations andoscillations of the structure are neg-ligible under live load and highwind conditions.

WIND ANALYSISThe structure was designed for

100 m.p.h. wind velocity and theforces were applied in accordancewith A.A.S.H.O. Specifications. Themain cables were assumed to becompletely encased by ice at thetime of maximum wind occurrence.The lateral flexibility of the maincable and suspenders caused a re-distribution of some load from themain cables to the stiffening girder.The analysis, based on deflection, re-

i0 PCI Journal

vealed that the lateral deflection atthe center of span was 11'/z in. forthe cables and 6Y4 in. for the stiffen-ing girder.

The small deflection value for thestiffening girder was due to its shal-low depth and large lateral rigidity.Longitudinal wind forces were trans-ferred from the deck to the maincables by tie rod connections at thecenter. The lateral wind load on thedeck was transferred to the lowercross beam of the tower by an artic-ulation joint. This joint allows theend of the stiffening girder to movelongitudinally and also to rotate ina horizontal and vertical plane aboutits center.

The end bearings of the structure(roller type) were centered underthe webs of the end unit. Shouldshrinkage and creep of the deck ex-ceed the assumed value, longitudinaladjustment of the bearing is pos-sible.

POST-TENSIONINGThe live load analysis as previous-

ly outlined revealed flexural stresseswhich are shown on Fig. 5 for thevarious loading and temperatureconditions. The girder was subjectto stress reversals with the maxi-mum ordinates occurring near thequarter point of the span. The stressreversals, though not of equal mag-nitude, required girder post-tension-ing forces such that the combinedstress for any condition of loadingwould remain within the followinglimits:

1. For Group I loading:+0 psi to +1800 psi

2. For Group III loading:+0 psi to +2250 psi

A 28 day ultimate concrete strengthof 4500 psi was sufficient for theabove stresses.

Several prestressing methods were

investigated including considerationof grouted or ungrouted tendons. Atthe outset, the investigation was in-fluenced by the fact that this struc-ture was the first of its kind andtherefore warranted close observa-tion over a sustained period of timeafter its completion. It was also ap-parent that the bridge could func-tion, in limited fashion, without anylongitudinal deck prestressing. Thedecision was therefore made to usestrands which were continuous fromone end of the structure to the other,ungrouted and so located that theycould be inspected, adjusted andeven replaced if the need ever arose.

The use of ungrouted tendons isnormally discouraged because ofcorrosion and lower ultimate loadcapacity. For this installation thegalvanized unbonded system wasjustified for the following reasons:

1. The crossing is in a dry beltand the girder detailing al-lowed for the circulation of airaround the strands. Moisture ispractically excluded from thestrands.

2. The galvanizing, even thoughsacrificial, is considered suffi-cient for protection. Strands canbe removed and replaced ifever required.

3. The post-tensioning system, isnot the primary load carryingsystem. Its failure would notcause collapse of the structure,only limit its live load capacity.

The absence of bond between, theprestressing strands and the con-crete girder forced consideration ofthe relative movements between thestrands and the girder. The maxi-mum movement, which occurrednear the point of inflection of thegirder, was calculated to be 21/4 in.Intermediate fixed or sliding sup-ports for the strands were consid-

April 1966 61

ered prohibitive because wearingforces would soon cause the strandsto fail and high friction would causemost of the prestressing force toconcentrate near the girder endswhere the force was least requiredby design. It was therefore decidedto adopt simple roller supports foreach of the strands, which supports,though expensive, resolved all theaforementioned problems.

The 19 wire, 1 1/s in. dia. galvan-ized strand, CCL system, was chosenfor the following reasons:

1. 34 strands in the top layer and36 strands in the lower layergave the desired distribution ofthe prestress force.

2. The end anchorage systemproved relatively compact andsimple.

3. Adjustment of tension or re-placement of strands was pos-sible if ever required.

The end blocks for the anchorages

were designed and reinforced forstresses as outlined by Guyon.4

Stressing of the strands was per-formed by calibrated center holejacks, one at each end of the strand.In this fashion, the value of frictioncould be determined. Because of theroller support system, the frictionwas found to be less than 3%.

MAIN CABLES

The maximum tension in the maincables occurred at the tower saddleswith the span fully loaded and atemperature of —50°F. This loadingproduced a tension of 4200 kips ineach main cable. In addition to thedirect tension, bending stresses alsooccurred in the cable when it ap-proached the curvature of the sad-dle or suspender clamps.

The above factors and the NorthAmerican practice of allocating axialunit stresses of 70 to 85 ksi for sus-pension cables, led to the selection

Fig. 9—Main Cable and Suspenders62 PCI Journal

Fig. 10—Main Cable and Suspenders

of 20-2'/s in. dia. galvanized bridgestrands for each main cable. A max-imum axial tension of 210 kips wastherefore imposed on each strandwhose ultimate load carrying ca-pacity was greater than 554 kips.The ratio of working to ultimateload was therefore substantially lessthan the permissible prestressed con-crete standard of 0.6.

It is interesting to note that forthe Merelbeke Concrete SuspensionBridge in. Belgium, 5 the main cableswere allocated an axial unit stress of112 ksi. The latter was more in linewith the allowable stress designa-tions by prestressed concrete codes.

Each 21/8 in. dia. strand consistedof galvanized wires and all wireswere full length with no splices. Allstrands were prestressed to one halfthe specified ultimate load and heldat this load for 4 hours. The loadwas then released to the averagedead load tension occurring at 30°F.

At this tension, the strand was meas-ured and marked for the locationsof sockets, saddles, suspenders, etc.The modulus of elasticity for eachstrand was obtained and the averagevalue of 24.2 ksi was above the min-imum of 23.5 ksi specified.

The 20 strands which made upeach cable (Figs. 9 and 10) wereleft separated and unwrapped, un-like the wrapped cable system em-ployed on most other suspensionbridges. The considerable economyachieved by this open system ap-peared justified because this PeaceRiver area is noted for its dry cli-mate and absence of industrialfumes. Sufficient space has been leftbetween the strands to permit thefuture application of a corrosion in-hibitor if ever required.

ROCK ANCHORAGESThe rock in the area of the anchor-

ages proved to be suitable for drilled

April 196663

rock anchors. The following pro-gram was performed to test its suit-ability.

1. Removal of overburden andwashing the surface so that avisual inspection for cracks andfissures could be made.

2. Drilling core logs for the pur-pose of locating fissures or find-ing strata that was weak inshear.

3. Performing pull-out tests onboth grouted and mechanicalanchors. The flexibility of therock was obtained by the useof dial indicators.

Evaluation of the above data bythe Department of Highways' Geol-ogists and the Consulting Engineersestablished the location of the an-chorages, the minimum depth belowthe surface where the anchor forcehad to be located, and confirmed thefeasibility of either mechanical an-chors or grouted anchors. For eitheralternative, the design was such thatthe anchor force was located nearthe base of the drilled hole before

final grouting, with the anchor act-ing in a manner similar to the endanchorages of a post tensioningstrand. Figs. 11 & 12 show thebasic layout of the anchorage hous-ing, anchor plates and the mechan-ical anchor. The Contractor, HansMordhorst Ltd., selected the me-chanical anchor alternative andmodified the anchor shoe arrange-ment.

Each anchor had to carry the loadof one cable strand, equal to a maxi-mum tension of 210 kips. The speci-fied minimum ultimate load for theanchor was 580 kips. The anchorwas stressed to a load of 290 kipsby the use of a center hole jack.This load was later relaxed to 230kips and the top of the anchorlocked to the anchor plate. Afterall the anchors were tensioned, theywere grouted for corrosion protec-tion. The above system, in essence,"prestressed" the band of rock be-tween the base of the hole and theanchor plates in the housing. Themaximum cable tension would re-

CONCRETEANCHORAGE -HOUSING 20 GALV.

BRIDGEFINISHED GRADE STRANDS

ROCK SURFACE ...

20 MECHANICAL ROCK ANCHORSSPLAYED - OUT BOTH VERTICALLYAND HORIZONTALLY.

Fig. 11—Anchorage Housing—Typical Section

64 PCI Journal

Fig. 12—Rock Anchorages

move most of the prestress, but aminimum of about 10%-15% of theanchor load would always remain.This system offered the followingadvantages:

1. No anchor load would be trans-ferred to the surface rock whichwould normally be fissured orwater bearing.

2. The anchor plates would re-main virtually motionless underall load conditions. (This is im-portant since it is desirable tokeep the point of zero move-ment as close to the tower aspossible.)

3. Fracturing of the surface rockby earthquake, erosion or ex-plosives would not endangerthe resistance of the anchorblock.

TOWERS

The use of concrete for the projectwas also extended to the tower con-

struction. Fig. 13 illustrates thegeometry and cross section of eachtower.

One of the most significant factorswhich affected the tower design wasthe saddle movement caused by tem-perature variation (+110°F to-50° F)and cable stretch under live load.The total movement for these con-ditions was 41/4 in. at the NorthTower and 41/z in. at the SouthTower. The placement of rollers un-der each saddle was considered asa means of reducing the bendingeffect on the towers caused by thesemovements. However there was nospecial advantage indicated in thisdetail since the tower section wasadequate in resisting the bendingstresses. Furthermore, with rollersunder the saddles, the free conditionat the top created a longer effectivecolumn length for the tower, whichhad disadvantages. Consequently,the saddles were fixed to each tower

April 196665

37'-0"^/ CABLES

P.I. EL. 1716.1'CABlE .SADDLE

1. 0••

TOP of DECK.1 TOWER --EL. 1615.0'

Fig. 13—Tower Details

leg.The idea of using a pinned tower

base in the longitudinal directionwas intriguing, but closer examina-tion from the viewpoint of towerconstruction and the erection of thedeck units, made this type of detailimpractical. However, to keep thestresses from the longitudinal move-ments as low as possible the longi-tudinal depth of the tower frommid height to the base was madeconstant.

The tower was assumed to havethe following properties for the de-sign analysis:

1. Transverse Direction—Rigidframe fixed at the base. Shaftsand cross girder having a vary-ing moment of inertia.

2. Longitudinal Direction.—Col-

umn fixed at the base, pinnedat top but subject to partialIongitudinal movement. Vary-ing moment of inertia frombase to top.

The critical design stresses in thetowers occurred from the axial load,longitudinal bending from the tem-perature movements and transversebending from lateral wind. Theabove, when summated and appliedto a rectangular cross section, pro-duced high governing stresses at onecorner of the tower leg section. Byusing the cruciform cross sectionfor the tower shafts, a more efficientdistribution of the combined stresseswas achieved. The cruciform shapewas not only more structurally effi-cient, but also more aestheticallyattractive. A 28 day ultimate con-

66PCI Journal

Fig. 14—Segme"Fal Unit on Low Bed

crete strength of 3000 psi was usedfor all the tower concrete.

Before the erection of the deckunits commenced, the saddles wereoffset from the tower center lines.These offsets, 161/2 in. for the northtower and 18 in. on the south tower,were required for the effects of theremoval of sag and stretch of thebackstays. The tower tops couldonly be deflected 21/z in., otherwiseoverstress would have occurred atthe base due to longitudinal bend-ing. A saddle jacking arrangementwhich consisted of hydraulic jacks,and a fixed greased plate under eachsaddle resolved the above problem.As erection of the units proceeded,the tower top offset was kept withinthe specified limits by this jackingsystem. After the full dead loadwas in place, each saddle was per-manently fastened to the tower byten 13/8 in. dia. anchor rods groutedinto preformed holes.

ERECTION OF SEGMENTAL UNITS

The erection of the segmentalconcrete deck units (each weighing90 tons) posed several problems notcommon to conventional steel sus-pension bridges. The most signifi-cant problem was the imposition ofthe very heavy deck unit load on acable system before the cables weresubstantially tensioned, thereby in-troducing the possibilities of localbending or "kinking" of the cablesin the region of the suspenderclamps.

Initially, it was felt that somepreloading of the cables would haveto be resorted to for the purposeof keeping the angle change at thesuspender clamps within tolerablelimits. The contractor devised twoalternate systems of erection, eachof which would have partially pre-loaded the cable prior to the erec-tion of the deck units. Finally, the

April 1966 67

contractor commissioned Dr. R. F.Hooley, Professor of Civil Engineer-ing, University of British Columbia,to conduct tests on a portion of themanufactured 2 1/8 in. dia. strand inorder to determine the safe anglechange the cable could withstand.Results of the tests indicated thatno preloading was necessary andvalues for the critical angle changeswere established for various cabletensions. Professor Hooley developeda revised erection system whichutilized the findings to advantage.The system of a cable traveller, slingcarriage and winching for erectingthe units was thus developed: Amodel was constructed to ensurethat the erection sequence adheredto the limits that were establishedin the cable testing program.

Each precast unit was first raisedin the casting bed, transferred to alow bed trailer and trucked to theriver side of the tower (Fig. 14).

From this position, a traveller, slingand two winches took over andswung the units to their final posi-tion over the river (Figs. 15 and 16) .

The traveller, a specially fabri-cated steel frame on four rubber-covered wheels, was mounted onthe permanent cable system of thebridge. Winch lines from each shoreof the river controlled the travellermovements along the cable. A steelsling, contoured to fit under the deckunits, was hung from the traveller.

The typical procedure for erect-ing each precast unit started withthe unit being lifted from the lowbed at the shore line of the river.Vertical dogs, bolted to the travellerand bearing against the main cableclamps, prevented the traveller fromslipping down the cable during thelifting operation. Two pairs of sus-penders immediately in advance ofthe unit were then pulled over andconnected to the sling. The unit was

Fig. 15—Erection of Segmental Units

PCI Journal68

Fig. 16—Erection of Segmental Units

then attached to both the sling linesand two slack pairs of suspenders.With the traveller in a stationaryposition, the sling lines were low-ered, the unit swung over like apendulum, and its load was thustransferred to the suspenders. Withthe traveller lines now unloaded butstill connected, the traveller dogswere retracted (by four hydraulicjacks mounted on the frame) andthen the traveller was moved downthe bridge cable to repeat the pen-dulum-like swingingoperationwhich eventually enabled the con-crete unit to be transferred to itsfinal suspender position on thebridge. In the center half of thespan, where the permanent bridgesuspenders became too short for theswinging operation, temporary sus-pender extensions were attached.

In this relatively simple and in-expensive fashion, six precast unitswere erected from the North bank

of the river, eleven from the Southbank, then ten from the North andthe final five from the South. Thissequence was necessary to keepwithin the permissible angular dis-tortion of the main cables. The clos-ing units, one at each end of thebridge, were then cast in place.

It is interesting to note that the32 precast deck units were placedin about 10 working days.

TEST PROGRAM

The actual behavior of this struc-ture, possessing a concrete stiffeninggirder which is unique was a matterof interest. A study program wastherefore undertaken during thesummer of 1965, some 9 monthsafter the bridge was opened totraffic, for the purpose of obtainingdata on the behavior of this struc-ture.

One aspect of the program wasthe comparison of the actual and

April 1966 69

theoretical behavior of the stiffen-ing girder when subject to the pas-sage of a 94 ton live load (two 47ton trucks side by side) over thespan.

The two trucks were positionedat 6 locations along the span. De-flection ordinates at the panel pointsfor each position were recorded. Thestructure was then analyzed for twoof the above locations.

Case 1—Live load at center lineof span.

Case 2—Live load at S8—Probableposition for maximum positivemoment in girder.

The same span constants whichwere adopted for the design werealso used for the above analysis.

Fig. 17 shows the theoretical andactual deflection curves for Case 2above. The elastic theory curve isalso shown as a matter of interest.

Examination of the results re-

vealed that the girder stiffness as-sumed for the design of the struc-ture is reasonably close to the actualin situ stiffness. The results for Case1 (not shown) revealed an evencloser agreement between the actualand theoretical deflections.

The stiffening girder moments forthe live load position of Case 2 areindicated on Fig. 18. Again, theelastic theory moments are shownfor comparison. The moment curvesclearly indicate the large differencein moments when this structure isanalyzed by either the deflection orelastic theory.

The maximum moment under the94 ton live load is about 70% of thedesign live load moment for thestiffening girder.

Another analysis has been per-formed for the 146 ton live loadspread over a length of 80 feet.This load will occur from a low bed

17 16 14 12 10 8 6 4 2 0 SPAN

1.0

0.8

0.6

0.4 /\ Kips

0.2 84/ \ 2p^ 84

0FT. +OFT.

\ 0.2

0.4

DEFLECTION \ /0.6

----ANALYSIS by ELASTIC THEORY /

ANALYSIS by DEFLECTION THEORY 0.8---ACTUAL

1.0

1 2 4 6,8 10 12 14 16 17 2SPAN 0

Fig. 17-94 Ton Test load—Stiffening Girder Defections70

PCI Journal

O ^!17 16 14 12 10 8 6 4 2 SPAN

Kips84

+l^K-FT O O K-FT.

4000MOMENTS 6000

---

200

—ANALYSIS by ELASTIC THEORY /

ANALYSIS by DEFLECTION THEORY 8000

100002 4 6 8 10 12 14 1617

CF SPAN 0

Fig. 18-94 ion Test Load—Stiffening Girder Moments

trailer carrying generators to thePortage Mountain Dam. The resultsof the analysis indicated that thisbridge can safely carry this load,whereas other two lane structures,designed for H25-S20 loading, wouldbe subject to overstress from the146 ton load.

CONCLUSIONSThe concrete box girder suspension

bridge provided an economical andaesthetically pleasing solution to theproblem of crossing the Peace Riverin the remote northern British Co-lumbia location. The relatively shortconstruction time of 10 months wasfurther proof of the feasibility ofthe structure. The use of reinforcedand prestressed concrete in a spanof 680 feet confirms that concreteis able to compete in the relativelylong span field which, to date, hasbeen monopolized by structual steel.

ACKNOWLEDGEMENTThe author wishes to express his

appreciation to the following per-sonnel who took part in the develop-ment and completion of this project.

Messrs. H. J. Barratt and L. Osi-pov, partners in the firm, providedguidance and assistance during thecourse of the project.

Col. H. H. Minshall first projectedthe concept of a segmental unit con-crete suspension bridge. Col. Min-shall currently holds a Canadianpatent on this concept.

Hon. P. A. Gaglardi, Minister ofHighways of British Columbia, andMr. J. Alton, Bridge Engineer forthe Department of Highways, gavethis project every encouragement.

Mr. Hans Mordhorst, the GeneralContractor, constructed the entirebridge.

Dr. R. F. Hooley, Professor ofCivil Engineering, University of

April 1966 71

British Columbia, assisted the con-tractor on all design problems per-taining to the erection system.

Mr. A. Hicks was the ResidentEngineer for Phillips, Barratt andPartners during construction.

REFERENCES

1. Steinman, D. B., "Deflection Theoryfor Suspension Bridges", TransactionsA.S.C.B. Volume 100, 1935.

2. Hardesty, S. and Wessman, H. E., "Pre-liminary Design of Suspension Bridges",Transactions A.S.C.E., Volume 104,1939.

3. Steinman, D. B., "Rigidity and Aerody-namic Stability of Suspension Bridges",Transactions A.S.C.E., Volume 110,1945.

4. Guyon, Y., Prestressed Concrete, Con-tractors Record Ltd., London.

5. Van De Pitte, Daniel, "Prestressed Con-crete Suspension Bridges", Proceedingson World Conference of PrestressedConcrete, 1957.

Presented at the Eleventh Annual Convention of the PrestressedConcrete Institute, Miami Beach, Florida, December 1965.

Discussion of this paper is invited. Please forward your discussion to PCI Headquartersbefore June 1 to permit publication in the October 1966 issue of the PCI JOURNAL.72 PCI Journal