On the Strains in the Interior of Beams .
By GEORGE BIDDELL AIRY , F.R.S. , Astronomer Royal .
Received November 6 , -Read December 11 , 1862 .
I iAVE long desired to possess a theory which should enable me to express and to compute numerically the actual strain or strains upon every point in the interior of a beam or girder , under circumstances analogous to those which occur in ordinary engineering applications , -partly for information on the amount of force actually sustained by the different particles of the cast or wrought iron in a solid beam , partly as a guide in the construction of lattice-bridges .
The memoirs and treatises on the theories of elasticity and strains , to which I have referred , have given me no assistance * .
I have therefore constructed a theory , in a form which ( I believe ) is new , which solves completely the problems that I had proposed to myself , and which , as I think , may , with due attention to details , be applied to all the cases that are likely to present themselves as interesting .
This theory , with some of its first applications , I ask leave to place before the Royal Society .
1 .
It is supposed , in the following investigations , that the beam consists of one lamina in a vertical plane , the idea of a solid beam being supplied by the conception of a multitude of such lamince side by side , all subject to similar strains , and therefore exerting no force one upon another .
It is also supposed that the thickness of the lamina is uniform , and that its form is rectangular , the depth of the beam being equal throughout : these suppositions are made only for the sake of simplicity , as there does not appear to be any difficulty of principle in applying the theory to cases not restricted by these conditions , although the complexity would be much increased .
It also appears necessary to suppose that the material of the beam yields equally , with equal forces , in different directions .
Another physical supposition , which appears to be necessary for complete solution of the problem , will be stated when we reach the discussion of the first instance .
2 .
It is to be remarked that our theory is not intended to take account of all the strains possible in a beam , but only of those which are introduced by the weight of the beam or its load in the position in which it is used .
A beam , whether of cast iron or of wrought iron , is , by the process of its manufacture , in most instances affected by permanent strains ; so that , while the lamina is lying on its flat side , some parts are ready to burst asunder , while others are severely compressed .
When the lamina is placed in a vertical plane , these accidental strains will be combined with the strains which are produced by the weight of the beam , &amp;c. ; nevertheless our attention will be confined strictly to the latter .
The algebraical expression of this idea is , that we do not want complete solutions of our differential equations ; we only want solutions which will satisfy those equations ; and among solutions which possess this property , we may have respect to the laws of pressure antecedently known from simpler investigations .
3 .
For the unit of force we shall use the weight of a unit of surface of the lamina ; but in writing the expressions , we shall omit the word " weight , " as no ambiguity can be produced by its absence .
For the unit of the force of compression , or of tension ( which is merely compression with changed sign , or negative compression ) , we must refer to such considerations as the following .
A force of tension is not a force acting in a single line ; it is a force acting in parallel or nearly parallel lines , with nearly constant magnitude over a considerable extent of surface .
In a large structure , like the Britannia Bridge for instance , on any space one inch broad there is a certain force of tension ; but on the neighbouring space of one inch broad there is the same force of tension , and so for each inch in a long succession there is sensibly the same force of tension .
The force of tension , acting on a certain breadth measured perpendicularly to the direction of tension , will therefore be proportional to that breadth , or will be equal to the weight of a surface or ribbon whose breadth is the breadth which sustains the action , and whose length varies with the magnitude of the tension .
That length is the proper measure of tension .
When the breadth subject to the action =1 ( the unit of linear measure ) , the amount of action is expressed simply by that length ; when the breadth has another value , the amount of action is the product of the value of breadth by the length which measures the tension .
The same remarks apply to the measure of compression .
4 .
We must now consider the effect of tension estimated in a direction inclined at an angle p to the direction of tension .
Suppose that a cut is made through the lamina , at right angles to the direction of tension , and that the effect of tension is to separate the sides of the cut .
And suppose the direction of tension to rotate in the plane of the lamina .
As the rotation proceeds , the tendency to open the cut diminishes , till , when p =90 ' , the tendency vanishes entirely .
But when ( p becomes greater than 90 ?
, the tendency to open the cut is restored , and when &lt; p=180 ?
, it is exactly as great as when , =0 .
, As p is further increased , the tendency diminishes by the same degrees , and vanishes for p=270 ?
; then increases till p= 360 ? .
It is never convTerted into a force of compression , and its changes are the same for positive and for negative changes of p. These considerations show that the effect must be represented by a formula containing only even powers of cos 'p .
und the following consideration will show that there will be only one term , multiplying cos2 'p .
When the tension acts at right angles to the cut , if t be the length which measures the tension , and if I be the length of a portion of the cut , the force which acts is the weight of the ribbon whose length is t and breadth I ; and is therefore = it .
But when the direction has rotated through &lt; p , the force acting obliquely on I is the weight of the ribbon whose length is t and breadth I. cos &lt; p , and is therefore =It. .
cos p. And this force is not normal to the cut , but makes the angle p with the normal ; and therefore the force which is normal to the cut , acting on the length 1 , is It .
cos pX cos p= it .
cos2 =IX t. cos2 'p .
Consequently the measure of the tension , at the angle 'p to the original tension , is t. cos2 p. The same theorem applies to compression .
5 .
We must now proceed to consider the coexistence of two or more forces of compression or extension .
There is no difficulty in conceiving that a plate of metal may at the same time be extended in one direction and compressed in another direction transversal to the former .
But on consideration it will be found equally easy to conceive that a plate of metal may sustain at the same time several forces of compression , or of extension , or of both .
It is easy to devise an apparatus which will produce these effects .
Such forces may exist in the strains of a beam ; and it is important to show that they can be included in a simple investigation .
The following theorem is now to be proved .
" Whatever be the number and directions of the forces of compression and extension , their combination may in all cases be represented by the combination of two forces at right angles , --these forces being sometimes both of compression , sometimes both of extension , sometimes one a force of extension and the other a force of compression , and generally unequal in magnitude .
" The following is the demonstration .
Suppose that there are forces of compression ( forces of extension being represented as negative forces of compression ) of magnitudes A1 , A2 , , &amp;c. , acting in directions which make angles aD , a , a &amp;c. with a fixed line .
Let us estimate the effect of their combination in a direction making any angle 4 , with the same line .
The angles between the directions of the several forces and this direction are respectively oa-4 , , - , , &amp;c. ; and therefore , by the last article , their effects in the direction 4 , are A1 .
cos2 ( a -4 ) , A , .
cos2 ( M24 ) , &amp;c. ; or A1 .
cos2 c1 .
cos2 , +2A1.cos 1 .
sin ac .
cos 4 .
sin +A1 .
sin2 c , .
sin2 4 , A,.cos2 a , .
cos2 2 .
+ 2A .
cos c2.sin .
sin , +A , .
sin2 , .
sin2 .
sin , &amp;c. ; the sum of which may be represented by &gt; ( A. cos2 a ) .
cos2 4+ 2(2A .
cos a. sin os ) .
cos 4 .
sin 4 , + ( A. sin2 ao ) .
sin2 4 , or a. cos24 + b. cos _ .
sin , +vc .
sin2 4 ; where a , 6 , c may have any magnitude and either sign .
And it is to be shown that we can find a force B acting at the angle P , , and a force C acting at the angle 13+90 ' , whose combination will produce the same effect .
Now the effect of these forces , by the theorem of last article , is B.cos2(P- ) +C .
cos2 ( +900A/ ) , or B. cos2 f. cos2 4 +2B.cos 3 .
sin 1 .
cos 4.s sin + B. sin2 P.sin2 +C .
sin2 3 .
cos2 Q -2C .
sin + .
cos t. cos C. sin. +C.cos2 P3 .
sin2 + .
Comparing this , term by term , with the former , B. cos2 3+C .
sin2 =a ; B. sin2 3+C .
cos2 3= c ; ( B-C ) .
sin2(3 =b .
The difference of the first and second equations gives ( B-C ) .
cos 23=a-c ; and the quotient of the third by this gives tan 2j3=- ; which always gives a possible value for 3 .
Then B-C= si or =-cos which is always possible .
And , by adding the first and second equations , B+C=a+c .
By the combination of B+C and B-C , B and C are found .
Thus all the elements may be found , for representing the effect of any number of forces of compression or extension , by the effect of two forces of compression or extension acting at right angles to each other .
Our succeeding investigations therefore will be confined to the consideration of two such forces acting at each point .
We are now in a state to proceed with the consideration of the strains in a beam .
6 .
In fig. 1 , Plate V. , let the parallelogram represent a beam , supported in any way , as for instance by having one end fixed into a wall , and subject to any force , as for instance the vertical reaction R of a support at distance h. If ER is negative , it will represent a weight hanging on the beam .
Conceive a line to pass in any curved or crooked direction , from the lower to the upper edge , dividing the beam into two parts , a near part and a distant part .
This division is to be understood merely as a line visible to the eye ; it is not to be contemplated as a mechanical separation ; for if it were such , the metal on one side could be considered as acting upon the metal on the other side only in the direction perpendicular to the separating line ; which action , in many cases ( as when the separating line is vertical ) , would obviously be incompetent to support the distant part of the beam .
The compressions and tensions , which we can suppose to exist while the continuity is mechanically uninterrupted , will suffice ( with or without other forces ) to support the distant part .
Now if the upper end of the curve terminates in the upper edge of the beam , conceive the curve to continue along that edge till it meets the upper angle at the end of the beam ; if it terminates in the vertical end of the beam , conceive it carried upwards till it meets the upper angle ; thus the special actions which sometimes operate in the limiting lines will be separated from those in the dividing curve .
Let r and s be the length and depth of the beam ; x the horizontal abscissa ( measured from o ) , and y the vertical ordinate ( measured from the lower edge ) 52 of any point of the curve .
At the first limit of the curve , the coordinates are z , 0 ; at the last , the coordinates are r , s. 7 .
The distant part of the beam is supported by the forces of compression ( this term , with negative values , including tensions ) across every part of the curve , combined with the reaction R. At the point whose coordinates are x , y , conceive that there is one force of compression B whose direction makes the angle I3 to the left side of y produced , and another force of compression C whose direction makes the angle P3+90 ?
to the left side of y produced .
And , in figure 2 , consider the actions of these on the small element 6s of the curve , or rather the actions on a portion of the lamina , including Ss .
Let 0 be the angle made by ts with y. The direction of the action of B makes with 4s the angle P +0 ; and therefore the breadth of the ribbon representing its action is 4sX sin ( + 0 ) , and its whole force is B. sX sin ( 3 + ) .
Resolving this in the directions of x and y , we have for the effects of B on the distant part of the beam , In the direction x , B. is X sin ( 3 +)X sin 13 , In the direction y , --B.s X sin ( P+ ) X cos 3 .
In like manner , the effects of C on the distant part of the beam are , In the direction x , C. sX sin ( P+90 ?
+O ) X sin ( P+900 ) , In the direction y , C. sX sin ( P+ 90 ?
+0 ) x cos ( + 90 ) .
Expanding the sine , we have , for the whole force in the direction x , { B. sin2 3+ C. cos2 p } .
cos 0 .
6s+{B .
cos. sin P3-C .
sin P. cos P3 } .
sin 0 .
S and for the whole force in the direction y , { -B .
sin p. cos P+C .
cos P. sin P } .
cos 0 .
s+ { -B .
cos2 3_-C .
sin2 3 } .
sin. $s .
But cos 0 .
s= y , sin 0 .
5s=8x .
And using for convenience the following letters , L =B .
sin2 3+C .
cos 2 , M=(B-C ) .
sin 3 .
cos 3 , Q= -B .
cos2 3C .
sin2 3 , we have for the whole forces on the element 4s , In the direction x , L. Gy+MAl .
x , In the direction y , A -M .
8y+Q .
x. It must be borne in mind that the force in direction x acts in a line whose vertical ordinate is y , and that the force in direction y acts in a line whose horizontal ordinate is X. 8 .
There is another force acting on this portion of the distant part , namely , the weight of the lamina included between the ordinates corresponding to x and x+^x ; which , estimated in the direction y , is -y .
4x , acting in a line whose horizontal ordinate is x. 53 And , besides these forces which act at every point of the curve , there is the reaction +R in the direction y , acting in a line whose horizontal ordinate is h. 94 We have now collected all the elements for the equations of equilibrium of the distant part of the beam , and we proceed to form those equations .
For by we shall put p &amp; x. The equations are as follows : First , equation for forces in x : Sdx .
( Lp+M)=O ... ... ... ... .
( 1 .
) Second , equation for forces in y : Sdx .
( -Mp+Q-y)+R=O ... ... ... ... ( 2 .
) Third , equation of moments : Jdx .
( Lip +My+Mxp-Q+y)-Rh=O ... . .
( 3 .
) It will be convenient at once to make y-Q-= 0 ; and the equations become dx .
( Lp +M)=0 ... ... ... ... .
( 4 .
) Sd .
( Mp +0)-=0 ... ... ... ... .
( 5 .
) Sdx .
( Lip +My+Mp+O)-Rh=0 ... ... ( 6 .
) 10 .
We shall now introduce a consideration which will prove singularly advantageous for the solution of these equations .
Referring to figure 3 , the equations which we have obtained apply to the curve abed of .
The same equations , mnutatis mutandis , apply to the curve abyd of : Hence the variations in those equations produced by passing from one of these curves to the other will =0 .
Now these variations are clearly such as are treated in the Calculus of Variations .
We may therefore form the variations of the equations according to the rules of the Calculus of Variations , and equate those variations to zero .
R and Rh will disappear .
11 .
The left side of equations ( 4 .
) , ( 5 .
) , ( 6 .
) , is in each case a function of x , y , ( L , M , and 0 depending on the position of the point in the lamina , and therefore being functions of x and y ) , and of no other differential coefficients .
Therefore the equation of variations in each case , in the usual language of the Calculus of Variations , will have the form Ndr -O .
Applying this in each instance we have ; For .
Sdx .
( ITp+M ) : dL dM d(P ) dL i dL N-dyP+ die ; PL ; -d d+dyP ; therefore dL dM dL dL dy2+ -dydyp or dM dL -dy .
d* ... ... ... ... For.jdx(Mp+40 ) : in the same manner , dO dM ... .
( 8 .
) dy-dr= IBEAM &amp; For Sdx(jIYp+My+Mp+O0 ) : dZL dM dO N= die p+L+y y+M+ die P+ x ; P=Ly+Mx ; d(P ) dl .
dL dM dM d -df-=y+@ pie +Lp+ dz ^+ d"yj+M ; therefore dL L dM dM dO dL dL .
dM dM dyYP+ip+ die +M + die * +d--dx Ydypy Lj ?
-d$^ WY -M=0 or ddM dl\ l dO dM\ .
dy-dc +S dy-d^ ) = ?
This equation , by -virtue of equations ( 7 .
) and ( 8 .
) , is identically true , and therefore adds nothing to our knowledge .
The information , then , that we have obtained from our process is comprised in the two equations dM dL7 dy-d ' ( * .
) dO dM dy-d ... ... . .
( 8 .
) From this it follows that L , M , 0 are the three partial differential equations of the second order of a function F of x and y , such that d2F dF d2F .
L== : M==r O=-dy2 M dxdy doc and we may substitute these symbols for L , M , 0 , in the equations of equilibrium of the distant part of the beam .
12 .
If it had been necessary to use expressions of the utmost possible generality , we must have said L= d2F d+ M 2F M d2F + where the forms of the functions qp and 4are arbitrary .
Suppose now that F is so d2F d2F d2F determined that the substitution of dy2 ddy , and d-2 for L , M , 0 will satisfy the equations ( 4 .
) , ( 5 .
) , ( 6 .
) in their entirety .
Then the substitution of p(y ) and 4(x ) alone must satisfy those equations deprived of their constant terms ; and therefore 9p(y ) and +(Zx ) may be multiplied to any degree , or different functions of the same character may be added to them .
These remarks clearly indicate that these functions represent accidental strains such as we have spoken of in article 2 , and they are therefore to be neglected .
We confine ourselves therefore to the terms d2 F dF dF L=d2 Mdxdy , ' 55 Making these substitutions , and restoring for pdyx its original expression die , the equations become the following : S(^)..(9 .
) ( d. dy2F + d. .
0 ( 1 .
) fd dxYdddy ( f( die * dy2 * y+d dxdY+d d +dx xd -RI=O ; *(1 .
) the integrals being taken from x , y=z , 0 , to x , y=r , s. Now d2F dF / dF\ d2F d2F / dF\ die die + dxdy d_dx-dy ; ddyd+ dd w7d 7T and the same symbols appear in the bracket of equation ( 11 .
) .
Hence the equations become *t.a ... ( 12 .
) .
( 13 .
) and J IY d( die ) + .
a d)}-Rh=O .
Integrating the quantities under the bracket by parts , the bracket becomes d+xdF-f die -+dx ) .
F dF dF dF d But d^y+.-dx=d(F ) .
The value of the bracket , therefore , is y +x -F ; and die ( ux bdy dx equation ( 11 .
) becomes fd{ ddy_ d ... .0 .
0(14 .
) Attaching the subscripts z , 0 and r , s to the symbols or brackets , to denote the values which the expressions assume when z , 0 or r , s are substituted for x , y , the equations finally become / d7P\ / dFv\ / -^ , =0U. .
( 15 .
) ( dF ) _(dF , .
r , &amp; kx 1-E=0 ... ... .(16 .
) dy +- } die +x -F -oRh=0. .
Vy ' -dx 1J , .
, rjrfy ( ; J0,. .
f. G.B AIRY ON THE STRAINS IN THE INTERIO1 OF BEAMS .
powers of x and y. Assume , therefore , F=S+ Tie +1Uy2+Vy9+Wy4+ &amp;c. , where S , rT , U , V , W , &amp;c. are functions of x ; then dF d= T+2Uy+3Vy2+4Wy3+ &amp;c. For r , s , the value of this is T , +2Ur , .
s+3Vr .
s2+4Wr , S+ .
&amp;c. For z , 0 , its value is T# .
The expression ( ( F ) will therefore contain the function T , , where z is absolutely arbitrary .
It is impossible that equation ( 15 .
) can subsist , except by making T-0O , and therefore T,.=0 , and generally T=O .
Again , omitting T , we find ( using the accents to indicate differential coefficients ) dF d=sF +UTyJ +v'y'+w'y4+ &amp;c. For r , s , the value of this is S:.+Ur .
s82+V .
s3+W .
s4 + &amp;c. For z , 0 , its value is For the same reason as before , S ' generally =0 .
Therefore if S have any value , it is a mere numerical constant ; and this will disappear in each of the equations ( 15 .
) , ( 16 .
) , ( 17 .
) ; and therefore it may be entirely omitted .
The expression for F will therefore be reduced to Uy2+Vy3+Wy4+ &amp;c. We shall hereafter show that ordinary investigations entitle us to assume that the expression for F will really be limited to the first two terms of this series , and that the powers of x will not be higher than the second ; and therefore we shall suppose F=(ax2+bx+c)y +(e 2+fx+g)y3 .
We can now proceed with instances .
15 .
Examnple 1 .
Suppose the beam to project from a wall , and to sustain no load except its own weight .
Here R=O ; and the three equations ( 15 .
) , ( 16 .
) , ( 17 .
) , with the last assumption for F , become ( 2ar2 ?
+2br+2c)s+(3r2+ 3fr+ 3y)s2= 0 ( 2ar +b)s2+(2er +f)s3 =0 , ( 3ar2+2br+c)s2 +(4er2+3fr+ 2)s3=0 .
Determining from these the values of b , c , g , we change the expression for F to the following : F { ( aXZ2+(-2ar-2ers-fs)x+ar2+2er2s+frs}y } l+ { e2+fx-er2-fr}y JK &amp; MTh 4 B. A ON T , IT IS I THE .
INTER:.IOR OBEAMS4 To determine the constants a , e , f , which remain , we must have recourse .
to:other eo3nT siderations .
16 .
If we suppose the beam cut through in a vertical line corresponding to abscissa x , and if we make the usual assumptions in regard to the horizontal forces acting betweeii the two parts and thus sustaining the .
moment of thei distant part , namely , that there is a neutral point in the centre of the depth-that on the upper side of this neutral point the forces are forces of tension , and on the lower side are forces 6of &lt; cmpression-and that these forces are proportional to the distances from the neutral .
point , with equal coefficients on both sides , then we can ascertain the horizontal iorce at every point .
But I remark that it appears to me that these suppositions involve a distinct hypothesis as to the physical structure of the material .
They seem toimpy that .
the actual extensions or compressions correspond exactly to the curvature .
of the edge of the lamina , and that the forces of elasticity so put into play correspond .
to the amount .
of extension or compression .
The experiments of Mr. W. H. BARLOW appear to modify this theory ; and it seems probable that , when duly followed into their mathematical consequences , they may require the introduction into the formula for F of other powers of y. Leaving this question open , I shall now proceed , on .
the usual assumptions , to com-pute the horizontal force at every point of the vertical division .
17 .
Let the horizontal force at elevation y , estimated* as compression , be represented by bo ) ; .
the force on the element 3y is the ribbon yx a. ( j-y ) ; its moment is yX yXC(--y)= and the entire moment is C % die , -^J)_l which , fromy==( } to y=s , is -^ .
The moment produced by the weight of the distant part of the bar is the product of its weight by the horizontal distance of its ca( Z s(rX)2 centre of gravity , or is ( r--2XsXr 2x= ) .
The equation of moments is theref e C3 S(r_X)2 ( r x)o2 fore , -12+ 2 =0 .
From this , c== 2 ; and the horizoniltal compression-force at elevation y=g6.(r-.X)2.-y ) ; or the horizontal compression-force on the element 18 .
But we have the means of expressing the same horizontal force in terms of F. For , in the last expressions of art .
7 , conceive the dividing line to be vertical ; that is , conceive 4x=0 , and s=by - , then we have for the compression-force on the element y in 'direction a , the expression L. y ; which , giving to L its value from the end of art .
11 .
d2F becomes -j2 By .
Comparing these two expressions , _y y(iy And , using the last formula of art .
14 , { 2ax2+(-4ar-4ers2fs)a+2a+42 } 6+ 2fs}S 6Y ) + { 6e+ +6fx6er2-fr } 2I 2r r2 Comparing the coefficients of y , ex2+f-+(-e -r)=- .
2+ , -The first '** * 1'2r 7 ' 2 2r2 From these ( see art .
7 ) , tan 2--L + ; C-B N L+Q C -sin 2/ cos 2p C+B=L-Q ; which give the numerical values of the three elements B , C , S of the strains at every point .
By means of these formulae , the numbers of Table I. ( end of the Memoir ) have been computed and the lines of pressure traced in Plate V. fig. 4 .
They give complete information on the nature and magnitude of the strains to which such a beam is subject .
20 .
Example 2 .
A beam of length 2r and depth s rests , at its two ends , freely on piers , and sustains no load except its own weight .
Assume , as before , F= ( a 2+ bx + Cy2+(e+ g)y3 , and remark that the distant pier exerts a reaction vertically upwards , of magnitude rs at distance 2r .
The three equations ( 15 .
) , ( 16 .
) , ( 17 .
) , taking the integrals from z , 0 to 2r , s , become ( 8a2+4br+2c)s +(12er2+6fr+3g)s2 =0 ; ( 4 r+6)s2 +( 4er +f)3 -rs =0 ; ( 12ar3+4br+c)s2 +(16er2+ 6fr+2g)s'--2r2s=O .
When from these we determine the values of b , c , g , and substitute them in the expression for F , it becomes F+ { ex2+fx+(-4er2-2fr)}y3 21 .
The horizontal pressure at any point of any vertical line across the beam at distance x will be found on the usual theory as follows .
The compression at any elevation y being represented , as in article 17 , by c. ( 2-y ) , the entire moment is , as in that cs3 article , - .
The moment produced by the weight of the distant part of the beam , whose length -is 2r-x , is ( ; and the moment produced by the reaction at the distant pier is -rs X ( 2r-x ) .
The equation of moment is therefore cS3 s(2r-x)-_ -~-2+ 2 -rsX(2r-x)=-0 , or cs2 ( 2r-x)2 2r(2r- , )2 12 ' 22 =0 or cs2 x(2r-x ) 12 2 6x.12rx From this , c= 2 - ; and the horizontal compression-force at elevation y 62 1 2rx Therefore , as in article 18 , d2F 6x-12rx / s dy2 2 2 , And , using the last formula of article 20 , { { 2a ; 9+ ( -8ar-8ers-2/ fs 8+)+(8ar2+16er2s+4frs- ) G}1 6-2 12rx(s + { 6ex2+6fx+(-24er--12fr)}y s6 212r Comparing the coefficients of y , 6e x+Gfx+(-24er2-12fr)=-x 2--S2 -x .
The 1 2r 24r2 24r w first term gives e= - ; the second givesf=- ; the third gives r =0 , which 2 , is identical .
Substituting these in the term independent of y , and comparing , 2ax'+ -8ar+ +x 8a'+ ?
--j= - .
2aw+ ( -8ar+ 4r ) , + ( 8a2 216r .
28r2 4r22 36 The first term gives a= s ; and on substituting this , the second and third comparisons become identical equations .
The evidence of correctness of theory is therefore satisd2F 6a2-12rx/ s\ factory ; and we may use dy2 22 .
From this we find F 6x'-l2rx ( Sy._y ) from which d2F 6X2_ 1rx / s\ = 6r2 / \ 2Y\ d2F 12x_ 12r(sy y2 6r M 12 r. y --s .
6 .
IY 1-Y =dxdy s.2 -S'T ' r -N= -2M I.s(1 ( 1- ) d2F 12 sy2 y3 ) As before , put v=- , w ; and suppose -=5 .
Then , omitting s , =-75. .
( 2-v ) .
( 1--2w ) ; N= 60.(1-v0).w.(l-w ) ; Q= w. ( 1-w ) .
( l-2w ) ; :h : .
B.'AIY OTE 'ST IS:3N TBE INTERIi OF BEAMS .
after which we may use the same formulae as before , namely , tanl2=3 L+ , N L+Q C-B =7L-sin 3cos 23 ' C+B =L-Q ; by means of which the numbers have been computed for Table II .
( end of the Memoir ) , and the lines have been traced that are exhibited in Plate VI .
fig. 5 .
23 .
There is one part of the pressures which it is matter of great interest to compute , namely , the pressures exerted on different parts of the end portion of the beam which rests on the pier .
It will be seen in figure 7 that this part is not free from pressure ; there are at every : point a large force of compression in one direction , and a large force of tensionin another direction .
And the circumstances of this part differ from those of any other vertical section of the beam in this respect , that there is no opposing force .
In all other sections , a thrust of compression on one side is met by a thrust of compression on the other side , and so for tension ; and though there may be a tendency to , crush or to disrupt , the particles of the metal , yet there is no great tendency to force a small sectional portion horizontally or vertically .
But on the end portion , where the forces of compression and tension are not so met , there are or may be great tendencies to farce that end portion horizontally or vertically .
We proceed now to investigate these tendencies .
24 .
First , for the horizontal pressure .
The force B ( which is estimated as a compression ) , acting in a direction which makes the angle 3 with the vertical , upon the element by ( as measured in the vertical direction ) or sin p.y ( as measured in the direction transverse to B ) , does really exert the pressure B sin P3 .
by in the direction of B , or the pressure B. sin/ 3 .
in in e , or B. sin2 3 .
/ y , in the horizontal direction .
Similarly , the force C exerts the pressure C. sin2 ( 3+90 ?
) .
y , or C. cos2'3 .
y , in the horizontal direction .
The entire horizontal force upon the element by is therefore ( B. sin2P+C .
cos2 3 ) .
, y=L .
Y= dy2 In the instance before us , of a beam resting on two piers , d2F 6x22rx / s\ and at the end of the beam , where x=2r , this quantity =0 : whatever be the value of y. The same applies where X= 0 .
There is no ' tendency therefore to bend or distort the end portion .
25 .
Secondly , for the vertical pressure .
The pressure B. sin ( 3 .
y in the direction of B , found in last article , will produce the pressure B. sin 3 .
cos 3 .
y in the direction vertically downwards .
Similarly , the force ' C will 'produce the pressure C. sin ( 3'+ 9P ?
) .
cos ( -p ?
900 ?
)Y ; SRAP~L~ P T1~ NT3~RQROF ?
t cfd iy = The } whole downwards pressure therefore on theele-mcent s io s ( B. si 3 .
cos+-p .
; sirin 9+ .
c.-eos 900+p3 .
by , or ( B-C ) .
si:n ( 3.eosS.9 , or M. i* *.12x-12r / Sy -y$\ ; V( ... . .
which in the present instance = .
2- " t- ) y. At the end of the beam , where x=2r , this =-- .
( sy--y2 ) .
tie .
' Let y'=s-y ( that is , let the ordinate be measured from the upper edge downwards ) ; then the downwards pressure on the element ' of the e4nd portion =--s ( sy ' -2 ) .
y ' .
Integrating this from the top downwards , we find for the pressure which a horizontal section of the end portion must sustain , f , r ( sy Y*y2 / 3_ At ie middle of the depth this = 2 ' ; at the base it =rs .
It appears therefore that every part of the end portion which rests upon the pier is subject to a very heavy pressure ( such as affects no other part of the beam ) , increasing from the top to the bottom , where it is equal to the weight of half the beam .
It was undoubtedly from a clear perception of the magnitude of this pressure ( though ; not reduced to the formule of mathematical investigation ) that Mr. ROBERT STEPHENSON , in the construction of the Britannia Bridge , was induced to insert the strd'ng end-frames in each of the tubes , aitthe places where they rest on their piers .
26 .
Example 3 .
A beam of length 2r and depth s rests , at its two ends , freely on piers , and carries a weight W at the distance a from the left-hand extremity .
For convenience , we will suppose a to be not greater than r. This will include every case , as the supposition a ' greater than r is the same as the 'supposition a less than " ' measured from the right-hand extremity , if a+a'=2r .
In examples 1 and 2 , we have selected a form for F which satisfied the equations ( 15 .
) , ( 16 .
) , ( 17 .
) , applying to F , and we have then shown that this form represents properly , the ' horizontal pressure determined from the ordinary theory .
In the present example , which is unsymmetrical and complicated , we shall find the form for F ( a discontinu'o'us !
form ) which represents the horizontal pressure as determined from the ordinary the'ory , and shall show that this ' form satisfies in all parts the equations ( 15 .
) , ( 16 .
) , ( 17 .
) .
2:7 .
The pressure upon the left-hand pier is , rs-+W *2ra ; and that upon the right-hand pier is rs+W ( The same value will be found if we consider the moment as produced by the weight of bar and , the reaction on the left side of x. ) Treating this as in article 17 , we find the horizontal compression-force at elevation y 6{(2r+W 2r -a ) ( d2F This , as in preceding instances , ought to equal -d2 ; and therefore F ought to equal dy { ^(2r+W .
2'S )2 } ( )64 This formula applies to any point of the part of the bar included between x=0 and x-a , which we shall call the " first part .
" For any point of the " second part , " or the part included between x=a and x=2r , there is no weight W on the right hand ; the forces producing moments are the weight sX(2r--x ) acting at distance 2r , and the reaction rs+W a at distance 2r-x ; the sum of their moments , estimated as compressing the upper part , is a 2r2r ( 2r-x)(rs+Wl 2 )=Wa+ ( rs-Wa)$ ; whence , as in article 17 , the horizontal compression-force at elevation y 6(2Wa / a\ / s\ which ought to equal d2 ; and therefore F ought to equal 6 2Wa +( ay3 sY2 s2{s+ ( 2r-Wr-w ax -S ) This formula applies to any point of the part of the bar included between x=-a and x-=2r , or to any point of the " second part .
" The function changes its form , or is discontinuous , when x passes the value a , the two formulxe , however , giving the same value for F when x=a .
We have now to ascertain whether the discontinuous function does in all parts satisfy the equations ( 15 .
) , ( 16 .
) , ( 17 .
) .
28 .
First , suppose the integrals to begin from a point z in the " first part .
" It is unnecessary to make an elaborate trial of equation ( 15 .
) , because , as our assumed value rsy dF ci2 sy dP for F contains the multiplier T--and therefore contains the multiplier 2-2y dy will necessarily vanish at both the limits for y ( namely y= 0 , y=s ) which enter into the formule of ( 15 .
) .
In regard to the other equations , the integrals must be taken by the formula of the " first part " from z , 0 , to a , s ; and by the formule of the " second part " from a , s , to 2r , s ; and the constant forces are +W at abscissa a and rs+W2)j at abscissa 2r .
For equation ( 16 .
) , in the first part " = 2r+W .
-2x } - ) ; which for , 0 , =0 , and for a , s , = ~{2r-a+ 2ra in the " secondpart " = s26{2r-22.* 12 A rs 4s which for a , s , and for 2r , s , s~il--2r The sum of the two portions of the integral will therefore be,.A x.2ra a2ra ) -{0+2r-2a+W .
2r 2r+2a+W'-2r-W2 rs2rW -rs W 2r -S rS .
S 2r -2r-a To this are to be added +W andrs+W-r , or -rs+W te s2rum is 0 2rI ' or - ; the sum is 02 Equation ( 16 .
) therefore is satisfied when z is in the " first part .
" dF For equation ( 17 .
) : omitting y dy ( because , as is explained above , it cannot produce any term ) , it will be found that in the " first part " dF 6f 2 sy\ X -S Z 6F=-4 )x dx )k4 which for z , 0 , =0 , and for a , s , =a And in the " second part , " dF 6=(-2Wa ) 3sy2 dx : s 8\6 4)/ ' which for a , s , 6 ( 2Wa 23 =si4s + at12}j and for 2r , s , 6 2fWa , S3 6{WT+4r2 } si2 The sum of the two portions of the integral will therefore be s0 f , 3 2Wa 2Wa 2{0+a s---a2+-s +4r =}2rs .
To this are to be added +Va and- .
( rs+W ) 2r , or 2s ; the sum is 0 .
Equation ( 17 .
) therefore is satisfied when z is in the " first part .
" 29 .
Second , suppose the integrals to begin from a point z in the " second part .
" As before , it is unnecessary to make a trial of equation ( 15 .
) , which is necessarily satisfied .
In regard to equations ( 16 .
) and ( 17 .
) , the integrals are only to be taken by the formule of the " second part " from z , 0 , to 2r , s ; and the only constant force is-(rs+Wz ) at abscissa 2r .
For equation ( 16 .
) , din the " second part " = j2r-Wr-2x-j ( 2 ?
-which s fxa6 a for z , 0 , =0 , and for 2r , s , =-2 -2r--W &gt; or rs+W-V .
To this is to be added -(rsa +W2j ; the sum is 0 .
Equation ( 16 .
) therefore is satisfied when z is in the c second part .
" dF 6( 2Wa - ) / s2\ For equation ( 17 .
) , d:--Fin the " second part " = ---* ( ) which for z , 0 , =0 , and for 2r , s , = --a+4 r2=+Wa+2rs .
To this is to be added -(rs+W2r , or -2rs-Wa ; the sum is 0 .
Equation ( 17 .
) therefore is satisfied when z is in the " second part .
" 30 .
It appears therefore that our equations ( 15 .
) , ( 16 .
) , ( 17 .
) are in all parts of this loaded bar satisfied by the discontinuous formula which we found for F ; and therefore that formula is to be adopted in the further calculations .
But different calculations must be made for the " first part " and the " second part .
" First Part , from x.=0 to x=a .
F = .
{ ( 2r+W.- ) ( Y ) L=s { ( 2r+W.2r a)( ?
) ; lE 1{ 2rrs f ( 2 2s ) 4N =.6 { 2r+W .
2r--a2 rm:= r , , , ; 2r -a s I &amp; 0=,.(8 2r3 r+2 ) y2 yS O=s *t 3E-2K_e S22 S3 ) Second Part , from x=a to x=2r .
FL6 { 2W +(2r-W )x } ( a SY2 ) 6 21 a x ?
s L. = .
+(%2r-W.)-(-xi ; M=_ -{2r The strains upon the beam are not at all affected by placing a weight upon its end ( supposed strong enough to resist distortion of form ) .
It appears probable , therefore , that the extreme changes of opposite character will be given , on the one hand , by placing the weight upon the centre of the beam 's length , or making a=r ; on the other r hand by placing the weight upon the centre of one half of the beam , or making a= .
We will proceed first with the formule for the case when the weight is upon the centre , or a=r .
It is unnecessary here to make calculations for the two segments of the beam , as the strains will be symmetrical with respect to the two extremities .
As before , s is taken =-5 .
Weight rs placed on the centre of the beam 's length .
L=-s26{ -2 2-s* 3-2 *3x x2 y)=-s .
75.v .
( 3-v).(l2w ) ; L= e{3rx }.2 =-S.= N -{3r--2x}.y.(s-y ) s= s. s --s s 60.-{ .
w.(l w ) ; Q= s.w.(1l-w ) .
( 1-2w ) ; from all which , as before , the general factor s may be omitted .
Proceeding now with the other case , or Weight rs placed on the centre of the first half of the beam 's length , the formulae for the " first part , " from x=O to x=a= , or from v=O to v=0'5 , will be 2L= 2 2-}(2-)- .
{ r-r2 } ( l-)=- .
7 .
, ( -v).(1-2 ; s6 c 12r s % ' " " " N= s-2 .
( 72-2x } y. ( s-y ) = s.s. -( = -s60.{0.}..(l-w ) ; Q -s. .
( I -w ) .
( 12w ) ; and those for the " second part , " from x=to x=2r , or from v=0'5 to v=20 , will be 6 3r",2 xL =+ 32r { -}=-S..(2-r ) ( 2+ ) .
( 1-Sy)=-s.75.(2-).( i-+v ) .
( l-2 ?
) ; 22 T. rrI N= ) =..= s , .
( . .
) .
( -- , w ) ; Q a= -s. .
w.(1-w).(1-2w ) ; from all -which the factor s may be omitted .
32 .
For all these cases , the same formulae as before are to be used in the ultimate calculations of the magnitudes and directions of the strains , namely , tan 23=L+Q N L+Q CB -sin 2 ' cos 2 .
' C+B =L-Q .
By means of these , the numbers have been computed for Table III .
, and Table IV .
parts 1 and 2 ( end of the Memoir ) , and the lines of figs. 6 and 7 , Plate VI .
have been traced .
33 .
It is worthy of remark that , in figures 4 and 5 , the lines representing the direction of thrust , and also those representing the direction of pull , are continuous ; but in figures 6 and 7 they are discontinuous , the two segments of each curve , at their meeting in the ordinate vertically below the weight , having different tangential directions .
This follows as an inevitable consequence of the assumption in art .
16 ; I think it probable that a hypothesis like that of Mr. W. H. BARLOW would remove the discontinuity .
An investigatiaon similar to that of art .
25 would show that , at these points , the transverse section of the beam must be sufficiently strong to support the weight by thrust ( if the weight is on the top of the beam ) , or by tension ( if the weight is carried by or attached to the bottom of the beam ) .
34 .
There are cases somewhat different from those already considered , whose importance and singularity of principle are such as to make them worthy of special notice .
In Mr. ROBERT STEPHENSON'S construction of the Britannia Bridge , the strength of the tubes was nearly doubled by the following admirable arrangement .
The junction of the ends of successive tubes , at their meeting on the piers , was effected , not while the two successive tubes rested on the bearings which they were finally to take , but while the distant end of one of the tubes was considerably elevated .
It is a problem of no great difficulty to ascertain what elevation ought to be given in order to reduce the maximum strains on the bridge to their smallest value ; when the best arrangement is made , the strains are reduced to one-half of their original value .
The singularity of the mathematical principle consists in this , that there is impressed on the end-frame of the tube or beam a strain of the nature of a couple , or ( as it is called in the preceding articles ) a moment .
Where there are three or more connected tubes , the middle tube , or each of the middle tubes , has such a moment-strain at each end ; but each of the external tubes has a moment-strain at one end only ( inasmuch as , at the land termination of the bridge , there are no means of applying such a strain ) .
There are therefore two different cases , requiring different investigations .
35 .
Take , first , the case of a middle tube in which a moment-strain is impressed on each end , the directions of the two strains ( supposed equal ) being opposed , so that both tend to raise the middle of the tube .
The pressures upon the two piers will not be disturbed , because the effects of the two strains upon the entire beam balance .
If now we consider theforces which act on .
the distant part of the beam ( using the lan68uage of art .
7 ) , we shall have to combine , with forces formerly recognized , the moment which acts on the distant end .
By the known laws of translation of the place of application of a moment , we may suppose this moment applied at the imaginary division of the bar .
Thus , at every vertical section of the bar , there is combined with the ordinary moment of strains a moment equal to that impressed on each end .
The most advantageous magnitude for this moment is evidently half the magnitude of moment at the beam 's centre , with opposite sign ; for if we use a smaller value we leave too much moment at the centre , and if we use a larger value we impress too great a straining moment at the junction above the pier .
36 .
Now in art .
21 we found , for the horizontal thrust in a point of any vertical section , 6X2rx ( -y ) .
As regards the variations of x , this is greatest when x=r , 6r2 Sr s\ and its value is then ( -y ) .
One half of this with changed sign , or + is now to be applied to the expression for horizontal thrust in every part of the beam 's length .
Hence the expression to be used for horizontal thrust or compression is 6x'--12rx+3r2 ( s _y ) and therefore F6x2-12rx+3r2 ( Syy2 It will be seen immediately that this quantity satisfies the equations ( 15 .
) and ( 16 .
) , the integrals being taken from z , 0 to 2r , s. But with regard to equation ( 17 .
) , we must consider that in the instance before us a moment is to be introduced which has not presented itself before , namely , the moment impressed on the distant end .
The value of that moment , which ( with the sign contemplated in forming equation ( 17 .
) ) is 3r2 / sy\ r2s -jdy..2--y2),.becomes +4 .
Hence equation ( 17 .
) becomes in this case { dF ?
dF_ F dF + dF F -2r2 rs+ 0 .
TYdy +$rs{d+d.-F--2rrs+ , 0 .
And , on making the substitutions , this equation is satisfied .
37 .
Therefore we are to adopt 6x2-1 2rx + 3r2 sy2 y3\ -s '\4 T 6 ' from which d 2F 6x2-12rx +3r s 3r2/ 1 2aX 2 L-dy2s2 2y= s2-2-+2 ; dCF 12x--12r sy_y2 ' Mdxdy= 2\2 2 =N=-2M = 2M 11-1- ) ; 6 r( x y.s _ 69 o=d2F 12 = ( 32 2 ) Q=y-O=s , O-3s+2)sS(l.(l -L Or , if v= , W=Y ) -=5 , and the multiplier s be omitted , L= 75i .
{ ( l ) , .
( -2w ) ; N=60.(1-v ) .
w.(1-- ) ; Q=w.(l-w).(1--2w ) .
Then NN L+Q tan 2P=L+ ; C--B=si o ; CBL-Q , by which the numbers for Table V. have been computed , and the curves of figure 8 , Plate VII .
have been drawn .
38 .
Take , secondly , the case of an end tube , on which a moment is impressed only at one end .
In this case , the effect of that moment is not balanced by a moment impressed at the other end , and must be balanced by an increase of pressure on the near pier ( at which the moment is impressed ) , and a decrease of pressure on the distant pier .
The value 4 of moment will be balanced by an increase of pressure -g on the near pier , and a decrease of pressure -8 on the distant pier .
Hence the pressure on the rs 7rs distant pier will be rs-8--8 .
From this ( as in art .
21 ) the moment produced by the weight of the distant part of the beam =s(2 , and the moment produced by 7rs the reaction of the distant pier is =--X ( 2r-x ) .
The equation of moments is now Cs s(2r-x- ) 7rs 1+ ( 2 X ( 2r-x)= 0 ; or cs3 , ^ .
f QQ 7r --1-+s(2r- ) r-2 8-=0 ; or C2_s(2r-x).( 8)=o From this , ( 6x12r ) ( w-0 sand the and the horizontal compression-force at elevation y ( 6x--12r)( 4 ) se 2 Therefore we are to take for trial ( 6x-12r)(-4)/ sy y. S '~2 46 7rs Remarking that the reaction of the distant pier =-8- , and that its moment upwards 7rs -=s -X2r , it will be found that this function satisfies equations ( 15 .
) , ( 16 .
) , ( 17 .
) .
39 .
Adopting therefore ( 6x 12r ) )2 y3 s2 ( 4 6)9 we have d2F ( 6x-12r ) .
27 L9 --A2 ( -2 2 =d~ddy~ 2 2= s ( 8 rs ( 1 s ) N=--2M 12r 9 x.y 1d2F _12 lsy2 Y3 2y 0= J ( s 8),.(8Y-2 ) OSY - .
--2 -gz~ C~i~y~2 -S2 SS 6S S-3 Q=y-O =s sY ( lYs).(1-^ ) And , with v-= , wo= , r=5 , s=1 , r 88L=-75.(2-v).v-4 ( 12w ) ; N= 60 -v ) .
w.(l-w ) ; Q= wt.(1-w ) .
( 1-2w ) ; NN L+Q tan23=L+Q ; C-B-i- ; C+B=L--Q ; by which the numbers for Table VI .
have been computed , and the curves of figure 9 have been drawn .
40 .
These instances will probably suffice as applications of the theory to the most important cases of practice , and as examples of the modifications on subordinate points which may be required in investigating strains where the forms or other circumstances are different from those considered here .
41 .
Perhaps useful information may be derived from the diagrams and tables of numbers for guiding the construction of Latticed Bridges .
Thus , in such cases as those of figures 5 , 6 , 7 , the upper and lower edges require great longitudinal strength in the middle of the beam 's length , but very little near the ends ; on the contrary , powerful lattice-work is required near the ends , but very little near the middle .
In the case of figure 8 these remarks require very considerable modification .
71

