20

7. The large expansion existing at the ends of the suspended spansmade some special expansion arrangement necessary in the connection be-tween the stringers and the floor beams. An ingenious arrangement toaccomplish this end was designed by Mr. Ralph Modjeski, AssistantEngineer, and is shown on Plate 43. It dispenses entirely with a longsliding surface and supports the end of the stringer perfectly.

8. As the end of the anchorage arm consists only of tension mem-

bers, it was difficult to make a satisfactory form of portal which wouldat once resist vibrations and have the substantial appearance whichseemed important in the most conspicuous part of the bridge. The re.suit was accomplished by making a stiff frame, entirely independent ofthe tension members, consisting of posts placed between the inclined eyebars of the end panels, these posts being connected by a stiff strut at thecenter, and the rectangular panel above divided by stiff diagonals. Thisstiff frame is attached to the pins of the truss, but the pin holes iji thestiff posts are made one inch larger than the diameters of the pins, so thatthe tensile strains should not be disturbed. r

Loads. —The views of the Chief Engineer have long been opposedto proportioning bridge superstructure for precise wheel-base loads, theseprecise loads being always those of a special locomotive, either actual orassumed, and subject to constant change. The superstructure of theMemphis Bridge is proportioned in accordance with the practice now fol-lowed by that engineer, the load per foot on a length of 20 ft. beingtaken as twice the load per foot on a length of 120 ft. and upwards, andthe load per foot being increased by one per cent for each foot in lengthless than 120 ft. The excess variation in web members, being the differ-ence between the strains produced by a moving load and by a fixed loadof equal intensity, is taken on a basis one half greater than the uniformmoving load. In the case of the Memphis Bridge the unit load wastaken at 4000 pounds per foot of track, and the wheel-base load on 20 ft.was consequently 8000 pounds per foot of track, while the stringer load,the panels being a little more than 28 ft., was taken [4000 X (1 + 0.92) =7680] at 7680 pounds per foot of track. These weights were used incalculating the live loads given on the strain sheets shown on Plates 50 to54, inclusive.

In the central span the dead load was assumed in the strain sheet at4000 pounds per lineal foot of truss, or 8000 pounds per lineal foot ofbridge, being just double the assumed live load. In point of fact, theactual weight of this span is 8300 pounds per lineal foot, exclusive of endposts which stand directly over the piers. This excess of weight amountsto less than four per cent, and is no greater than the difference in weight

THE MEMPHIS BRIDGE.

-between the kind of floor used on this bridge to accommodate highwaytraffic and the usual railroad floor.

In the intermediate or suspended span the dead load was assumed at5400 pounds per lineal foot. In point of fact it is 5660 pounds, or alittle less than five per cent in excess of the calfeulated weight, the r actualdifference being almost exactly the same as in the central span.

In the anchorage and cantilever arms no fixed rate of dead load wasassumed, but the strain sheet was made by assuming the estimatedweights of the different-members as concentrated at the separate panelpoints, these concentrations, of course, being greatest nearest the piers.In every case the weight of a tension member was supposed to be carriedat the upper end of that-member, and the weight of a compression mem-ber to be carried at the lower end of that member, this, of course, onlvapplying to the resultant portion of that weight which moved in thedirection of the axis of the member. The panel point loads are given onthe strain sheets.

The upper lateral system is'proportioned to resist a horizontal press-ure of 300 pounds per lineal foot of bridge. The bottom lateral systemis proportioned to resist a horizontal pressure of 600 pounds per linealfoot. In both instances the whole horizontal pressure is treated as amoving load. This differs from the Usual practice; but the allowancemade for moving loads is inadequate rather than too great, as it is amongthe possibilities that when half of a span is exposed to a wind pressurein one direction, the other half may be exposed to a wind pressure inprecisely the opposite direction. r

No allowance is made in the strain sheets for the disturbances of thedistribution of weight by wind pressure. A simple calculation showedthis to be unnecessary. The lateral pressure on the top chord will exer-cise its greatest disturbing effects in the central span. Taking this press-ure at 300 pounds per foot, multiplying it by the depth of the truss, anddividing it by the distance between truss centers, we have a resultant of800 pounds as the possible increase of weight on one truss due to this cause,this being 20 per cent of the estimated dead load and 13 per cent of thetotal estimated load. If we take a wind pressure of 400 pounds per linealfoot applied on a train of maximum weight at an average elevation of 8ft. above the rails, the effect of this wind pressure is to move the centerof gravity of the moving load 0.75 ft. from the center of the track; asthe center of the track is 15 ft. from the center of each truss, this in-creases the moving load carried by one truss five per cent and diminishesthat carried by the other five per cent, thus actually increasing the weightthrown on one truss 100 pounds. Taking these two disturbances to-

gether, the total extreme weight which can be thrown on one trassamounts to 900 pounds per lineal foot, or 15 per cent above the assumedcalculations.

No addition was made to the chord sections to provide for lateralstrains. The assumed lateral force on the bottom chord is 600 poundsper lineal foot; the assumed total vreight carried by each truss is 6000pounds per lineal foot; the depth of the trusses is 2.6 times the distancebetween centers of trusses, so that the strain thrown into the chord by thelateral system is equal to 26 per cent of that thrown in by weight. Inthe top chord the effects of the lateral strain would be one half thisamount, or 13 per cent.

As the effect of the disturbance in weight by wind pressure is to in-crease the tension in the leeward bottom chord 15 per cent, while theeffects of the strain in the lateral system are to increase the tension in thischord by 26 per cent, the total effect is an increase of tension of 41 percent in the leeward chord and a corresponding reduction of tension onthe windward chord. In the top chord the effects are much less. "Withthe unit strain allowed, these strains are well within safe practice.

Proportioning of Materials. —The entire superstructure of theMemphis Bridge is of steel, and it was all worked as steel, the rivet holesbeing drilled in all principal members and punched and reamed in thelighter members.

The tension members were proportioned on the basis of allowing thedead load to produce a strain of 20 000 pounds per square inch, and thelive load a strain of 10 000 pounds per square inch. In the case of thecentral Span, where the dead load was twice the live load, this corre-sponded to 15 000 pounds total strain per square inch, this being thegreatest tensile strain.

The compression members were proportioned on a somewhat arbi-trary basis. They were generally designed so as to make them of sym-metrical section, and almost always so as to make them symmetrical aboutthe line dividing the least transverse dimension. No distinction wasmade between live and dead loads in proportioning compression mem-bers. A maximum strain of 14 000 pounds per square inch was allowedon the chords and other large compression members where the length didnot exceed 16 times the least transverse dimension, this strain beingreduced 750 pounds for each additional unit of length. In long compres-sion members the maximum length was limited to 30 times the leasttransverse dimension, and the strains limited to 6000 pounds per squareinch, this amount being increased by 200 pounds for each unit by whichthe length is decreased.