Q. Q. 1: ' Express the following linear programming problem in the standard form: Maximize Subjectto Z = 3x 1 + 2 ~ + 5 ~ 2x l -3x 2 ~ 3 , Xj+ 2X;; + 3 ~ ~ 5 , 3x 1 + ~ ~ 2 , Sol Solution. Here Xl and ~ are restricted to be non-negative, while X3 is unrestricted. Let us express X3 as x 3 ' - X3" where X3' > 0 and X3" > O. Thus the above problem can be written as Maximize ~ 3 ~ 5 ~ 2 where Introducing slack variables, the standard form is Maximize Subject to =3 =5, =2, 30 .BC-2.4 Q. 2 : Express the following LP. problem in the standard matrix form Maximize Z= 4Xl + + x 3 , Subject to 2x 1
3x 1 =8,
Solution. The problem can be expressed in the standard form and then represented in the matrixform as follows : Maximize Z = 4x 1 + + + OX 4 + OXs )blem = (4,2,6,0, 0) subject to 2xl + + =6, =8, 6x 1 - + X3 + Xs = 10, xl> These constraints can be expressed as Xl 3 2 -1 x2 4 0 0 x3 (! -:-4 0 x4 Xs , Thus the gjven problem in the matrix form is Maximize Z=cx Subjectto Ax = b, . IC-2.4 31 BC-2.4 where c = (4,2,6,0,0). and A = ! 6 -4 1 0 1 Q.3: A chemical company produces two products, X and Y. Each unit of product X requires 3 hours on operation I and 4 hours on operation II, while each unit of product Y requires 4 hours on operation I and 5 hours on operation II Total available time for operations I and II is 20 hours and 26 hours respectively. The production of each unit of product Y also results in two units of a by-product Z at no extra cost. Product X sells at profit of Rs. 1 D/unit, while Y sells at profit of Rs. 20/unit. By product Z brings a unit profit of Rs. 6 if sold ; in case it cannot be sold, the \ destruction cost is Rs. 4/unit. Forecasts indicate that not more than 5 units of Z can be sold. Formulate the problem so that the quantities of X and Yto be produced, keeping Z in mind, so that the profit earned is maximum. Solution. Formulation of L.P. Model Step 1 : The key decision to be made is to determine the number of units of product X, Y and Z to be produced. Step 2 : Let the number of units of product X, Yand Z produced be x1J X 3
where X. =number of units of Z produced = number ofunits ofZsold + numberofunitsofZdestroyed =Xa + x 4 (say). Step 3: Feasible alternatives are sets of values Xa, and x 4 where. Step 4 : Objective is to maximize the profit. Objective function (profit function) for products X and Y is linear because their profits (Rs. 10/unit and Rs. 20/unit) are 32 8C-2.4 c constants irrespective ofthe number of units produced. A graph between the total profit and quantity produced will be a straight line. However, a similar graph for product Z is non-linear since it has slope + for first part, while a slope of 4 for the second. However, it is piece-wise linear, since it is linear in the regions (0, 5) and (5, Y#). Thus splitting X. into parts, viz. the number of units of Z sold (><s) and number of units of Z destroyed (x 4 ) makes the objective function for product Z also linear. uct X nit of Total '. The tZat I I it. By I, the Zcan Jced, its of ><s. 1) for ) are Thus the objective function is maximize Z =1OX 1 + 20x 2 Step 5 : Constraints are On the time available on operation I : 3x 1 + 20, On the time available on operation II : 4x 1 + 26 On the number of units of product Z sold: X3 < 5, On the number of units of product Z produced: 2X2 =X3 +x 4 or +X3 +x 4 = 0 +6x 3 - 4x 4 Q4: A firm produces an alloy having the following specifications: (i) Specific gravity < 0.98, 1:2 (ii) Chromium .8%, (iii) Melting point 2: 450 0 C. Raw material A, Band C having the properties shown in the table can be used to make the alloy. Table Property Properties of raw material A B C Specific gravity 0.92 0.97 1.04 Chromium 7% 13% 16% Melting point 440C 490C 480C Cost of the various raw materials per unit ton are: Rs. 90 for A, Rs. 280 for Band Rs. 40 for C. Formulate the problem so that the proportions in which A, Band C be used to obtain an alloy of desired properties while the cost of raw materials is minimum. ,33 BC-2,4 -2.4 Solution. Formulation of L.P. Model Step 1 : Key decision to be made is how much (percentage) of raw materials A, B and C be used for making the alloy. linear Programming & Sensitivity Analysis. Step 2: Letthe percentage contents of A, Band C be X 1 , ~ and X3 respectively. Step 3 : Feasible alternatives are sets of values of Xl' where j =1,2, and 3. Step 4 : Objective is to minimizethe cost i.e. minimize Z =90x1 +280x2+ 40x3 Step 5 : Constraints are imposed bythe specifications required for the alloy. They are 0.92x1 +0.97x2 + 1.04x3:$; 0.98 . 7x1 + 13x3 + 1x3:$;8 440x1 + 490x2 +480x3 :$; 450, \ x1 + x2 +x3 = 100 Q. 5 : Mohan-Meaking Brewries Ltd. has two bottling plants, one located at Solan and the other at Mohan Nagar. Each plant produces three drinks, named A, Band C respectively. The numbers of bottles produced per day areas follows: Product Plant at Solan{S) Mohan Nagar (M) A 1,500 1,500 B 3,000 1,000 C 2,000 5,000 A market survey indicates that during the month of April, there will be a demand of 20,000 bottles of A, 40,000 bottles of Band 44,000 bottles of C. The operating costs per day for plants at Solan and Mohan Nagar are 100 and 400 monetary units. Formulate the problem so that each plant be run in April so as to minimize the production cost, while still meeting the market demand. 34 BC2.4 6.,B and dC dof ting tary 1ize
Solution: Formulation of L.P. Model Step 1 : Key decision is to determine the number of days for which each plant must be run in April. Step 2 : Let the plant at Solan and Mohar, Nagar be run for XI and X:z days. Step 3 : Feasible alternatives are sets of values of 0 and X:z 0 which meet the objective. Step 4: Objective is to minimize the production cost Le., minimizeZ = 100x 1 + 400x 2
Step 5 : Constraints are on the demand. i.e., forA, 1,500x 1 + 1,500x 2
for B, 3,000 Xl + 1,000X:z 40,000 andfor C, 2,000Xl + 5,000 x 2 44,000 cf6: Examples on Mixed Constraints Linear Programming Probem Use the graphical method to solve the following LP problem Maximize Z =7x 1 + 3x 2 Subject to the constraints Xl + XI + X:zs4 Osx 1 s5/2 OSX:zs3/2 and X 1
Solution. Rewriting the given constraints as follows. Graph each constraint byfirsttreating it as a linear equation. Then use the inequality condition of each constraint to make the feasible region as shown in graph below. x2 3 3/2 Xl + X2:::;;1 3 4 l:::;;I' x2 :::;;1 5/2 '3/2 35 8C-2.4 c 5 , XI"'" 5/2 3 ( 2 Xl +2x2 =3 2 3 4 5 6 Graphical Solution of LP Problem The coordinates ofthe extreme points ofthefeasible region are :A= (5/2,1/4),8 = (5/2,3/2), and C = (0, 3/2) The value of the objective function at each of these extreme points is as follows: Objective function value Extreme Point Coordinates (x}.x2) Z= 7xI +3x2 A (5/2: ~ ) 7(512)+3(1/4) = 7314 B (5/2), (3/2) 7(5/2) + 3(3/2) = 22 C (0,3/2) 7(0) + 3(3/2) = 9/2 The maximum value of the objective function Z = 22 occurs at the extreme point 8(5/2, 3/2). Hence the optimal solution to the given lP problem is : X 1 = 5/2, X 2 = 3/2 and Max Z = 22. 0.7: Use two-phae simplex method to maximize subject to the constraints: 2 ~ + x 2 ::; 1, x 1 +4X2 ;:::0 and Xl' x 2 ;:::0 Solution. Introducing a slack variable S1;::: 0, a surplus variable S2;::: 0, and an artificial variable A 1 ;::: in the constraints of the loP.P., an initial basic feasible solution is : Sl = 1 and A,=6 with /2as the basis matrix. Phase 1. The objective function ofthe auxiliary loP.P. is to maximize. z* = - Al Using now simplex algorithm to the auxiliary linear programming problem, the iterative simplex tables are: 36 BC-2.4 08. SoIL '4), B= NS: value 73/4 :22 )/2 e point (2 =3/2 ariable : 1 and m, the cB 0 -1 08. Initial Iteration. Introduce Yz and drop Y3. 0 0 0 -1 YB xB Yl Y2 Y3 Y4 Y5 Y1 1 2 I 1 0 0 .l'l 6 1 4 0 -1 1 Z - -1 0 1 -1 zi Cj Z (: -6) -1 -4 0 1 0 Since Z1 - c 1 and Z2 are negative. we choose the most negative ofthese, viz . -4. The corresponding column vector zz-c z enters the basis. Further, sincemin {XBi ,Yi2>O} Yi2 is I, which occurs for element yi 2 Y31eaves the basic. Final Iteration. Optimum solution CB YB xB )'1 Y2 0 Y2 1 '2 1 -1 Y5 2 -7 0
Z(=-2) 7 0 Y3 Y4 Ys 1 0 0 -4 -1 1 4 1 0 Since all (z j - c j) :;:: 0, an optimum basic feasible solution to the auxiliary L.P.P. is obtained. But max, z* < 0 and an artificial variable is in the basis at a positive level. Therefore, the original L.P.P. does not possess any feasible solution. Use two-phase simplex method to Maximize z = 5x 1 4X1 + 3X3 subject to the constraints: 2X1 + X 2 6X3 = 20, 6x 1 + 5x 2 + 1OX 3 s:; 76 8x 1 - 3x 2 + 6X3 s:; 50, X 1 ' x 2 ' x 3 :;:: 0 Solution. Introducing slack variables and, the L.P.P. in its standard form is: Maximize z = 5x 1 -4x 1 + 3x3subjectto the constraints : 2X1 + X 2 - 6X3 =20, 6x 1 + 5x l + 1OX 3 + S1 = 76 8x 1 -3x l + 6X3 + S2 = 50,X 1 , Xl' X 3 ,S1' Sl:;::O IC-2.4 37 BC-2.4
i ~ ,I The set of constraints in matrix form is -6 0 0 5 10 1 0 (: -3 6 0 xl Xz x3 ~ [ m s1 s2 Since, there does not exist columns of an identity matrix which can be considered for the starting basis matrix. We add requisite identity column( s) so as to complete the identity basis matrix. That is, insert the identity column [1 0 0] as the new column Y 6 Clearly, this amounts to adding an artificial variable in the 1 51 constraint. Now, an initial basic feasible solution isA 1 = 20, S1 = 7 and s2=50. Phase 1. The objective function of the auxiliary L.P.P. is z* =-A 1 . The iterative simplex tables for the auxiliary L. P. P. are: Initial Iteration. Introduce Y1 and drop Y5 ~ , 0 0 0 0 -1 Co YO Xo YI Y2 Y3 Y4 Y5 Y6 -1 Y6 20 2 1 - 0 0 1 0 Y4 7 5 10 1 0 0 0 Y5 50 8 -3 0 1 0 Z*(=-20) -2 -1 6 0 0 0 Since Z1 - c 1 is negative, the column vector Y1 enters the basis. Further, since min. : XBi ,Yil >o}= 50 Ysleavesthebasis. The element Y31 (= 8}becomesthe { Yil 8 leading element. 38 BC-2.4 -1 idered nplete e new -aint. Y6 o o o flih.: First Iteration. Introduce Y2and drop Ys' CB YB xB YI Y2 Y3 Y4 Ys Y6 -1 Y6 15/2 0 7/4 -15/2 0 -1/4 1 0 Y4 77/2 0 29/4 11/2 1 -3/4 0 0 Yt 25/4 1 -3/8 .% 0 1/8 0 Z* (=-15/2) 0 -7/4 15/2 0 1/4 / 0 Here Z2 - C z is the only negative Zj - c j This indicates the Y2 enters the basis. XB' } 15/2 Also min. _, 'Yi2 >0 =-- Suggests that Y6 must leave the basis, { Yi2 . 7/4 thereby Y12 ( -;) becomes the leading element. cB YB xB YI Y2 Y3 Y4 Ys - Y6 0 Y2 30n 0 1 -3017 0 -117 417 0 Y4 52n \ 0 0 25717 - 1 2n . -29n 0 YI 55n 1 0 -617 0 1114 3/14 Z* (= 0) 0 0 0 0 0 ., 1 Since, all Z ,- c;::: 0 an optimum solution to the auxiliary L.P.P. has been reached. Furthermore, it is apparent from the table that no artificial variable appears in the basis. Phase 2. Now, we consider the actual costs associated with the original variables. So, the objective function is Z =5x 1 - 4X2 + 3X3 + 0&1 + 0&2 39 BC-2.4 C-2.4 The iterative simplex table for this phase is : 5 -4 3 0 0 cB YB xB Yl Y2 Y3 Y4 Ys -4 Y2 3017 0 1 -30/7 0 -117 0 Y4 5217 0 0 25617 1 217 5 . Yl 5517 1 0 -617 0 1114 Z* (= 155/7) 0 0 6917 0 13/4 Since all Zj - c j ~ 0 an optimum basic feasible solution has been reached. Hence, an optimum basic feasible solution to the given L.P.P is x 1 =55/7, x 2 = 3017, X3 =0; maximumz= 155/7. Q. 9: Use penalty (or Big M) method to Z=6Xl + 4X2 Maximize subject to the constraints: Is the solution unique? If not, give two different solutions. Solution. Introducting slack variabless 1 0, S2 ~ 0 and surplus variable S3. ~ 0 in the constraints, the standard form ofL.P.P. is: Maximize z = 6x 1 +4X2 + 0.S1 + 0.S2 + 0.S3 2X1 + 3Xz + S1 = 30, 3x 1 +2X2+ s2=24,x 1 +X 2 -S 3 = 3, Clearly, we do not have a initial (starting) basic feasible solution. So, we introduce an artificial variable A1 ~ 0 in the third constraint. Then, an obvious initial basic feasible solution is S1 = 30, S2 = 24 andA 1 =3 Now, corresponding to the basic variables 51' S2 and, the matrix Y =B 1 A (where B =I the identity matrix) and net evaluationsz J - cj G=1,2,3,4,5,6) are computed, where c a =(0, 0, -M). 40 BC-2.4 . o Ys -117 217 1/14 1314 nce,an 'aints: ) in the ltroduce ial basic lereB = I ld, where BC-2A The iterative simplex tables are: Initial Iteration, Introduce Yl and drop Ys' cB 0 0 -M . YB Y3 Y4 Y6 xB 30 24 3 Z(=-3M) --.. >1 2 3 1 -M--6 - 4 0 Y2 Y3 3 1 2 0 1 0
0 __ L-_ . 0 Y4 0 1 0 0
0 Ys 0 0 -1 M '-- -M Y6 0 0 1 0 - In the above table Zl-Cl and z:,rc z are negative. Among these two Z l -G 1 gives the most negative value (since M is very large), therefore Yl enters the basis. Further, min. 'Yil >0 This indicatesy,leaves the basis and y" becomes the leading element. Since Ys corresponds to artificial variable A l , we drop it from the objective function. First Iteration. Introduce Y5 and drop Y4' () YB Y3 I T xB 24 0 1 1 i Y4 0 I 2 0 Y4
, 15 0 , -1 0 1 3 6 L
I 3 Z (= 18) I
0 2
0 0 0 0 I -1 -6 . {XBi o} 15 Since zs- C s <0, Y3 enters the basis, Further, min, - ..-, YiS > =-3 ' hs Y4'eaves the basis and YZ5 becomes the leading element. 41 BC-2A Final Iteration. Optimum Solution. CB YB I XB Y3 ---i 14 ' 0 0 5 Y5 , 6 8 YJ. i z (= 48) I Q.11 , Y4 Ys l -2/3
0 .SOll 1 113 1/3 0 I 2 0 r Y3 1 0 0 0 : I 0 -1/3 1 2/3 0 0 I Since all Zj - 2 O,an optimum solution has been reached. Thus, an optimum c J basic feasible solution to the given L.P.P. is x 1 =8 and X2.=O with max, Z = 48. Alternative Solution. It is evident from the net evaluations of the optimum table that the net evaluation corresponding to non-basic variable x 2 is zero. This is an indication that the current solution is not unique and an alternative solution exists. Thus we can bring Yz into the basis in place of Y3 or Ys' The resulting new basic feasible solution will also be an optimum one. Therefore, by introducing Yz into the basis in place ofY3' the new optimum simplex table is
Ys I 4 Yz 42/5 0 1 3/5 -2/5 o o Ys I : 115 115 1 3915 : 6 Y1 12/5 I' -2/5 315 o
Z (= 48) i 0 0 0 2 l O i We observe from this table that a different basic feasible solution results, but the optimum value of Z remains the same. Hence, the two basic (optimum) feasible solutions to the given L.P.P. are: x 1 = [8, 0] and x z =[12/5, 4215] 42 BC-2.4 ---- n :tble ) an ists. asic )Iex ", the ible Q.to: Use penalty (or Big M) method to Maximize z = Xl + 2X2 + 3X3 - x 4 subject to the constraints: Xl + 2X2 +3X3= 15, 2x l +x 2 + 20, Xl + +X3 +X4 = 10, Xli . . Solution. From the constraints of the problem, we observe that we do not have requisite identity column to get the starting B as an identity matrix. So, we introduce artificial variables Al ;::: 0 and ;::: 0 in the first and second constraints respectively. An initial basic feasible solution. then, is . = 15,A a =20 and X 4 =10 Now, corresponding to basic variables A 1 , A2 and X the basis matrix Y= B- 1 A and 4 , the net evaluations ZrCj (j = 1, 2; 3, 4, 5, 6) are computed, where c a (-M -M -1). The iterative simplex tables are: Initial Iteration. Introduce Ys and drop Y5' I CB -M -M -1 YB Y6 Ys Y4 I z xB Yl 15 1 20 2 I 10 1 I-35M -10 1-3M-2 Y2 2 1 2 -3M-4 Y3 3 5 1 -8M-4 Y4 0 0 1 0 Ys 0 1 0 0 I Y6 1 0 0 0 i' Since the most negative ZrCj ( =zs-c s ) corresponds to Ys. it enters the basis. Further X } 20 . min. { :; ,Yi3 >0 =5' the current basis vector Y51eaves the basis and Y2S becomes the leading element. As Yscorresponds to an artificial variable A 2 , we drop the Yscolumn from subsequent simplex tables. . First Iteration. Introduction Ya and drop Ys' CB -M 3 -I YB XB Y6 3 Ys 4 Y4 z -3M+6 -115 215 3/5 M 2 5 5. Y2 7/5 115 9/5 -7M 1 5 5 Y3 '0 1 o o . Y4 o o o Y6 o o o 43 BC-2.4 .4 -- Clearly, Z2-C2 is the only negative z(c j (since M is very large) and hence Y2
enters the basis. Further, Min. {XB.i ,Yil >o} corresponds to Y6' SO'Y6leaves the basis and Y12 Yn becomes the leading element. Again, Ys corresponds to artificial variable A1 Solut and therefore we drop the artificial column Y6 in the subsequent tables. Second Iteration. Introduce Y1 and drop Y6 . CB XB YB Y3 Y4 Y2 Y1 -117 2 1517 1 0 0 Y2 -- 0 3 2517 317 0 1 Y3 . 617 -\ 1517 \ 0 0 Y4 i---- 1---- ----- -17 9017 0 0 Z 0 I Clearly, Z1-C1<O and, therefore, Y1 enters the basis. Further, min. ,Yn >o} corresponds to. Y4 So, Y41eaves the basis andY31 YI1 becomes the leading element. Final Iteration. Optimum Solution. r-- cB YB xB Yl Y2 Y3 Y4 2 Y2 15/6 0 1 0 116 3 Y3 15/6 0 0 1 -3/6 1 i---- Y1 z -.--. 15/6 15
1
0 0 0 7/6
0 0 1 Since, all zr<1 are positive, therefore, an optimum basic feasible solution has been attained. Thus, an optimum solution to the given L.P.P. is 44 . BC-2.4 hence Y2 5and Y12 IriableA 1
Y4 11
-=:J 7/6 ! las been
Use the graphical method to solve the following lPP. Minimize z=-x1+2xz subjectto the constraints: - Xl + 3x z ::; 10, Xl +x 2 ::; 6 x 1 -x 2 ::;2 and x 1 ::::0, x 2 ::::O Solution. The constraints are re-written in the intercept form. Thus, we write XI + x2 <1 -10 10/3-' 6 6 -, and .::L+ x2 2 2 First, we treat these inequalities as equations and graph them as straight lines. Now, considering the inequalities, shade the feasible region for each constraint. The common region OABCD gives the solution space of the lPP. It may be noted that we have considered the feasible region only in the first quadrant of the Cartesian axis, as :::: 0 and . Xl 3 Fig. XIT3XZ 10 The coordinates ofthe extreme points are: 0= (0, O),A= (2, 0), B= (4, 2), C=(2. 4)and D=. (0, The z-value corresponding to extreme points are: Extreme point x z ) 0 (0.0) 0 A (2.0) -2 .... Minimum B (4,2) . 0 C (2.4) 6 20 D (0, In - 3 45 BC-2.4 BC-2.4
The minimum value of z occurs at the extreme point A (2,0). Hence, the optimum solution ofthe LPP is : x 1 = 2, x 2 = 0 and minimum z =-2 v6 12 Use the graphic,al method to solve the following LPP : So Maximize z = 2X1 + 3x2 subjecUo the constraints:
and Solution. To graph the given inequalities, we first treat them as equations X 1 + x 2 = 30, X 1 - X 2 = 0, X 2 = 3, = 20 and X 2 = 12 and plot each of these equations as straight lines. We use the inequality condition of each constraint to plot the corresponding feasible region. The common region ABCDE satisfying all the inequalities (feaSible region is shown in shaded Fig.) X2 Xl =20 40 10 20 40 Fig. The coordinates of the extreme points of the feasible region are: A= (3, 3), B = (20, 3), C= (20,10),0 = (18,1?), and E.= (12,12). The z-values corresponding to extreme points are: S . Extreme point (X1 + x 2 ) z = + 3x 2 A (3,3) 15 B (20,3) 49 C (20,10) 70 o (18,12) 72 maximum E (12,12) 60 The maximum value of z occurs at the extreme point 0 (18, 12). Hence, the optimum solution is : x 1 =18, x 2 . = 12 and maximum z =72. 46 8C-2.4 the optimum =30, !ht lines. We ling feasible sible region ence, the BC-2.4 Formulate the dual of the following linear programming problem: (j3. Maximize z = + subject to the constraints: + 15,5x 1 + . Solution. Standard Primal. Introducing slack variables S1 0 and S2 0, the standard linear programming problem is : Maximize z = 5x 1 + + 0.S1 + 0.S2 subjectto the constraints : 3x., + + S1+ 0.S2 =15,5x 1 + 2x:z + 0.S1 + = S1' Dual. Let w 1 and w 2 be the dual variables corresponding to the primal constraints. Then, the dual problem will be : Minimize z *=15Y11. + lOltZ subjecttothe constraints: i 3Y11. +5ltZ 2ltZ i I
=> WI 20 and W2 0."] W 1 and W 2 unrestricted (redundant). The dual variables "w 1 and W 2 unrestricted" are dominated by WI 0 and 0 . \ Eliminating redundancy, the restricted variables are W1 0 and W 2 O. Writethe dual ofthe following linear programming problem: Minimize z = 3x 1 - + 4><a subject tothe constraints:
+ 4x., + Solution. Standard Primal. Introducing surplus variables S1 0, S2 0, S4 9, O,and a slack variable S3 O,the standard form of LPP is : Minimize z = 3x 1 - 2x:z + 4><a + 0.S1 + 0.S4 + 0.S5 subjectto the constraints: 3x., + 5x., + 4><a- S1 = 7 6x., + + = 4 = 10 + = 3 4x., + = S5 = 2 S1' S2' S3' S4' S5 0 47 BC-2.4
Dual. If Wi G= 1, 2, 3, 4, 5) are the dual variables corresponding to the five primal constraints in the given order, the dual ofthe standard primal L.P.P. is :. . Maximize z*=7wt +4W:2 +1OY13 + 3w 4 + 2w s subjecttotheconstraints: 3wt + 6w 2 + 7Y13 +w4 + 4w s:::;5 5wt +W:2 - 214 3-2w4 +7ws :::;-2 -wt :::;0, - W:2 :::;0, YI3 :::;0, -w4 :::;0, -ws Wj G=1,2,3,4,5) are unrestricted in sign. Eliminating redundancy, the dual variables are : 0, W 2 2 0, 0, W 4 20 and Ws 2 O. Note. In the primal, if we multiply the third constraint throughout by -1, the dual variable W3 will be 20 instead of O. Q.15 Use dual simplex method tosolvethefol/owing L.P.P. : Minimize z=3xl +x2 subjecttotheconstraints: 1, , cB
YB xB Yt Y2 Y3 Y4 0 Y3 -1 -1 -1 1 0 ... 0 Y4 -2 -2 -3 0 1 (Zj-Cj) 0 3 1 0 0 L-._ Since all z j- c j 2 0 and X (=S1) <0, X (= S2) <0; an optimum but infeasible solution B1 l2 has been attained. In order to obtain a feasible optimum solution, we select a basis vector to leave the basis and non-basis vector to enter the basis as follows: Since, min. {XBI' X B1 <O}= min{-1, -2}= -2 (=X S2 ); basis vectorY4willieve the basis. Further, since the non-basisvector Yz will enter the basis. 48 I/e primal
the dual Y4 o o solution ta basis :isis. First Iteration. Drop Y3 and enter Y4' cn xn I J1 Y4 ! Y2 Yn Y3 -113 -1/3 0 0 1 -1/3 Y3 -1 2/3 2/3 1 -113 0 Y2 i I -2/3 7/3 0 0 113 I (Zj-Cj) I I I I . . . l' Intheabovetable. rn;n {xBi,xm (=xBi), and c Z'-C' . 1 {7/3 1/3} ( ) J J 'Ylj<O =max. __'_.'_ =_1 z4- 4 I { -1/3. -1/3 Y14 . Y 1j Thus Y31eaves the current basis and non-basi,s Y4 enters the basis. Final Iteration. Optimum Solution. I I I It is now apparent that since all Z j - C j 0 and also all X B1 0, an optimum basic . feasible solution has been reached. Hence, the optimum basic feasible solution to the given L.P.P; is: X 1 =Oand X 2 = 1 with minimumz= max. z*=-(-1)= 1. CB YB xB YI Y2 Y3 Y4 0 Y4 t 1 I 0 -3 1 I -1 Y2 1 \ 1 I -1 0 (= -1) 2 0 1 0 . 49 BC-2.4 BC"2.4 UNIT - 3 : NON-LINEAR PROGRAMMING (CLASSICAL METHODS) SECTION A MULTIPLE CHOICE QUESTIONS 1. In a non-linear programming problem a. the objective function is non-linear b. oneor more oftheconstraints have non-linear relationship. c. both (a) and (b) d. none of the above 2. Inthe context of constrained optimization with equally constraints a. a non-linear programming problem is solved using Lagrange multipliers b. the necessary conditions becomes sufficient for a maximum (or minimum) of the objective function if the objective function is concave(convex) and the side constraints are equations c. the sufficient conditions can beverified bythe bordered Hessian matrix d. all ofthe above 3. Which ofthefollowing is not correct? a. The method of Lagrange multipliers is a symmetric way of generating the \ necessary conditions for a stationary point. b. Any non-linear programming problem can be solved by the usual Simplex Method c. In some cases it is not generally possible to solve the equations resulting from the set of necessary conditions, explicitly for thevalues of x and Oils. d. The matrix HB is called the bordered Hessian matrix 4. In non-linear programming problem a. a local minimum of a strictly convex function. on a convex set is a unique global minimum ofthatfunction b. a local maximum of a concave function on a convex set is not a global maximum of that function c. if the constraints of an NLPP are equations optimum solution cannot be obtained by using Lagrange multipliers d. none ofthe above. BC-2.4 51 BC-2.4 SECTION D Nide LONG ANSWER TYPE QUESTION lave
Q.1. Derive the optimal solution from the Kuhn-Tucker conditions for the problem. Minimize Z = 2xI +3x2 xt- subject to the condition Xl +3x2 s6,5xI . . Solution. {(x)= 2xI + 3x 2 conditions are -Xl-3x2 2 hI Cx) =xI +3x2 -6, h (x)=5xl +2x2-1O For f (x) to be a minimum, Kuhn Tucker's necessary and sufficient conditions are that f (x) be convex and H (x), h 2 (x) .in x. It can be seen that these conditions are satisfied. The Lagrangian function is L(x,sA)= f(x)-AI [hl(x)+StJ-A2 [h 2 (x)+si] The K T conditions are (1 ) (i) 2-2xl-Al-5Az =0 (ii) 3-4x2 31.,1 -2Az =0 . non z a (2) linear I (i) Al (xl +3x2 -6) = 0 . . ' i (ii) Az( 5x l +2x2 10)=0 (3) (i) Xl +3X2 -6s0 (ii) 5Xl +2X2 -lOs0 (4) Al and Az There are 4 case to be dealt with now. Case 1., Al =O,Az then (1) (i) and (ii) give xl =1,x2 . This solution satisfies all the four K T conditions. 4 C-2.4 57 BC-2.4 18 20 , ,(i) g'ivesxI +3X2 -6:::0 } . xl--x2- Case 2. Ap!:O"-2 *0, then (2) . " , .. -13 -13 (ii) gives 5xl +2x2 -10::: 0 Al = negative Butthis solution does not satisfy the condition (4) and hence discarded. " (2) (ii) gives5xI +2x2 -10=0} 89 91 Case 3. Al ==0"-2 *0, (1) (? .. x 1 =90,x 2 =36 , ,'(ll) glves3-4x2 -2A2 =0 This solution violates conditon (3) and hence discarded , (2) (i) gives xl +3x2 -6=0 } Case4. Al *0"-2 =0 (1) (? =0 :. (ll) glves3-4x2-3Al =0 This solution does not satisfy condition (3) and, therefore, discarded Hence optimal solution is Xl = 1, x2 = Min Z == 17 Ans. 4 8 Minimize f(x) xl +x2 2,4,2x} +x2 25 and x120,x2 20 using Kuhn-Tucker conditions. ' Solu. f(x) = xl +xi,h 1 (x)=Xl +x2 -4,h 2 (x)= 2x l =x2 -5 We are to minimizef(x) subject to the constraints hI (x) ;::: 0, h z (x) ;::: 0 and x;::: 0 . The Kuhn-Tucke( conditions for the minimization NLPP are given as Ji(X)=Al hi (x) +"-2 hl(x) 0=1,2) Al hi (x)=O (i = 1,2) h i (x)20 ' (i = 1,2) A1 20 (i = 1,2) Qf(x) . oh i (x) where .Ii (x) =:= -;::::- and hi = a 0=1.2) and 1.. 1 , 1..2 are Lagrangian U,A,j , 'Xi multipiers. The conditions (1) (i) 58 BC-2,4 18 20 . (i) gives xl +3X2 -6=0 } :. xl =-x2 = Case2. Al *0"-2 *0, then (2)'. . 13 13 (ii) gives 5xI + 2x2 -10=0' Al = negative But this solution does not satisfy the condition (4) and hence discarded . . ' '. (2) (ii) gives5xI +2x2 -10=0} 89 91 Case 3. 1.,1=0"-2*0' (1) ... XI=90,x2=36 . . (ll) glves3-4x2 -21.,2 =0 This solution violates conditon (3) and hence discarded . (2) (0 gives xl +3x2 -6=0 } Case4. Apt: "-2 =0 (1) (i) gives2-2xI- A l =0 :. (ii) gives3-4x2-3AI =0 This solution does not satisfy condition (3) and, therefore, discarded Hence optimal solution is Xl =1, x2 = Min Z =.!2 Ans. 4 8 Minimize f(x)xf ,s.t.xI +x2 and Xl using \ Kuhn-Tucker conditions . . Here f(x) =Xf +x2 -4,h 2 (x)=2xI =x2 -5 We are to minimizef(x) Solu. subjeCt to the constraints hI (x) 2 0, h z (x) 2 0 and x 20. The Kuhn-Tucker conditions for the minimization NLPP are given as It(x) = Al (x) +"-2 hl (x) 0= 1,2) Al hi (x)=O (i =1,2) (i = 1,2) hi (x);;::
(i = 1,2) i ol'(x) I' oh (x) where fi (x) =: _(J_ and hj == 0= 1,2) and A 1 , A2 are Lagrangian . Oxi. . oXi multipiers. The conditions are (1 ) (i) BC2.4 58 s (2) . (i) 1...1 (Xl + = 0 nizef(x) Inditions BC-2.4 (ii) 1...1 (2X1 + X 2 - 5) =0 (3) (i) Xl (ii) 2x 1 + (4) 1...1 Case 1. 1...1 *0, =0 :. (1) (i) and (ii) give Xl = =0 . This does not satisfy condition (3), hencediscarded :. xl =I,x2 3 xl +x" -4=0 } Case2. 1...1 *0,1... 2 *0. :. (2) (i) and (ii) give , 2x1 +x2 -5=0 Ivz= negative. Here solution violates condition (4) :. discarded (I) (i) 2xI =2"-2 1 :. xl =2,x2 =1 Case3. 1...1 = 0, 1... 2 *0 (li) 2'(2 =A2 'Ivz =2 (2)(ii)2xl +x2 -5=0 This sOlution does not satisfy condition (3) :. discarded (1) (i) 2xI ='-1 } Case4. 1...1 *0, 1...2 = 0:.. 2x2 =A1 . ... Xl =2,X2 =2,Al ' (2) (I) xl +x2 -4=0 . This solution satisfies all the conditions . . Hence optimal solution is Xl =2, = 2, Min Z =8 Q Solve the following non-linear programming problem, using the method of V' Lagrangian multipliers: 2 2 MinimizeZ = x/ + X 2 + X3 subjecttotheconstraints Xl + + 3x 3 = 2,5xl + 2x 2 + X3 = 5 Solu. The Lagrangian function is '\ 2 2 2 . L(x,J\.)=xl +x2 +x3 -At (XI +x2 +3x3-2)-A2 (5xl +2x2 +x3- 5 ) The stationary point (><"0, AD) is given bythe following conditions . . Solution of these simultaneous equations is given by BC-2.4 59 ( ) ( 37 16 13) (2 7) and 'A ('A 'A) Xo = xl> x2 >x3 = 46 > 46 > 46 . = 1> 2 = 23 > 23 The bordered Hessian Matrix at this point is given by 0 o 1 2 3 0 o 5 2 1 ... Hg= 1 5 2 0 0 1 2 0 2 0 3 1 0 0 2 Here n = 3, m = 2; :. n - m = 1 and 2m + 1 = 5. For both (i) and (ii) we have to calculate the principia minor ofthe order 5. This means we have to check HoB only. IH 81=460>O and has the sign of (-1t,Le., (_1)2, Le., positive and hence minimum point. ><0 is a minimum point. .. Zmin= 0.a57 Q.4 Using the method of Lagrangian multipliers, solve the non-linear programming problem Maximize Z =6x 1 + aXz X 1 2 - Xz 2 Subject to the constraints 4 ~ + 3Xz =16 3x 1 +5Xz =15 60 . BC-2A aveto only. Iming Solution. 2 2 !(x)= 6X l + 8X 2- X l -x2. hI (x) =4xl +3x2 -16 h 2 (x)=3Xl +5x2 -15 The Lagrangian function is L(x, "'-)=!(X)-A l h l (x)-"--2 h2 (x) = xl + 8x 2 - xl 2 - x2 2 -AI (4xl +3x2 -16)-"--2 (3x1 +5x2 -15) The stationary point (xo, "'0) isthus given by the following conditions. 8L =0=6- 2xl-4Al - 3A2; 8L =8- 2x2 3Al-5"--2 =0 &1 &2 8L 8L " 8Al =-(4x1 +3x2 16)=0, 8A2 =-(3xl +5x2 -15)=0 " The solution ofthesesimultaneous equations gives xO=(xl,X2)=(35111,12111) The bordered Hessian matrix at this solution is given by Here m = 2, n = 2, we have to calculate the principal minor ofthe order 2m + 1 = 5, here it is to be 4 only. We check determinants as below: all al2 a13 al4 all ani" -\a 2l a22 a23 a24 a12 12 all a a22 I 'la 2l a23 and a31 a32 a33 a34 a41 a42 a43 a44 Iani, 1":21 a22 a31 aj2 a33 We see thatthe firstthree determinants are Oand the last 4x4 matrix. I.e., HOB =121 >0, (no alternating sign). Since HoB = 121 >0, xoisa minimum point. HenceZ m,n =2S1 61 BC-2.4 -2.4 Q.S Use the method of Lagrangian multipliers to solve the following non-linear programming problem. . Minimize Z= 2xf x ~ +3xf +lOxl +9x2 +6x3 -100 . Subjectto the constraints Xl +x2 +x3 = 20, Xl , x2 , x3 ;::: 0 Solution. The Lagrangian function is 222 L(Xl> X 2,X3,A)= 2x l +x2 + 3x 3 +10xI + 8x 2 + 6x 3-lOO-A(Xl +x2 +x3 -20) The necessary conditions for the stationary points are Q S Solving the above equations, we have Xl = 5, x2 = 11x3 +4, A =30 The solution of these simultaneous equations yields the stationary point Xo =(xl>x2,x3)=(5,1l,4) and A:::;130 The sufficient condition for the stationary point to be a minimum is that the minors ~ and L\4 be both negative. 0 1 1 1 0 1 1 o . 1 4 0 1 4 o <2 <0 and 83 :::; Now 84= 1 0 2 0 1 0 2' 1 0 0 6 The above values of each element ofthe determinant can be easily calculated out. Since 6: 3 and L\4 are both negative Hence the solution ~ = (5, 11,4) is minimum with min. Z = 281. Determine the relative maximum and relative minimum values of the function. f(x)= 12x;Z45x;t+40x) 5 ' Solution. It is a function of one variable. .. f(x)=O=60 (x 4 -3x 3 3 + 2x 2 ) x=O,x= 1andx=2 62 SC-2.4 lon-linear 20)
Ie minors ded out. iOn. BC-2.4 f (x) = 60 (4x 3 - 9,c +4x) At x =1, r (x) =-60 < 0, r it implies that f (x) is relative max. at x = 1 and hence maximum f(x=1) = 12. At x =2, f" (x) = 240 > 0, i.e., f(x) is relative minimum at x = 2, Min f(x=2) = -11. f"'(x)=60(12x 2 -18x+4)=240atx=O At x =0, r (x) =0this implies that we have to investigate the next derivative Since f" (x):;t:O at x=O, the point x=O is an inflexion point. Q.7. Find the relative minimum or maximum (if any) of the function . . 2 2 f(x) =xlf 2x2 +xI x2 -xl -x2 Solution. Here function f(x/consists of more than one variable. The necessary condition is BL -=O=1+x2 -2xI ax l BL . .-.=O=2+xl- 2x2 Qx2 . The simultaneous linear equations yield XQ = ,%) , For the sufficient condition, the Hessian matrix is obtained as a 2 f a 2 f H(XO)=\ Qx2 1 a 2 f QxlQx2 a 2 f =[-2 1] 1-2 Qx2 Qxl
The principal minors of H(xa) are: 1-21 =-2Bf1d . =4-1 =3[changingsignasper(-1)9{Here 1-21 means determinant 1 a11 1 } HenceH(xa) is negative definite and hence f(x) is relative . (4 3) 7 maximumat=s's .Zmax="3 63 aG-2.4 O. 8. Obtain the' necessary and sufficient conditions for the optimum solution of the following NLPP MinimizeZ =f(Xl,X2)=3i XI +l +2e X2 +5 subjecttotheconstraintsx1+x2=7 and Xl! > < : z ~ 0 Solution. Let us introduce a new differentiable Lagrangian function L(x1 ><:Z. A.)defined by L(x 1 .x A.) = f(x 1 x 2 ) - A. (Xl + < : z ~ 7) where A. is the Lagrangian multiplier. 2 The necessary conditions for minimum off(Xl , xJ are given by (1) Q.10. (2) Solu' (3) (or, ><:z = 7 - Xl) 7 12 X1 Hencefrom (1). (2) and (3) we have 6ix1 +1 =2tf2 +5 =2e -.,\1 +5 =2e 1 log3+(2x1 +1)=12-x or Xl =-(U-log3) Or 3' '. 1 X2=? -xl =-(lO+log3) 3 0.9. Find the solution ofth.e following problem Minimize f (x. y) = kx 1 y.2 subject to the conditions 2 g (Xi y) = ~ + y -a = 0 Solution. The Lagrange function 'is 1 k -1 -2 '1 (2 2 2) L( X1,X2'1I.)= x y. +"" x +y -a 64 !ution of the The necessary condition for the minimum of f(x, y) gives dL 0 kx- 2 -2 2 '1 0 -= =- 'y + XI\.= +x 2 =7and dxl dL =0=":2kx- 1 y-3 +2XA=0 , dX2' ,)defined by If the signs of minors L\3' L\4' As. etc, are alternatively +ve and ve, the stationary point is a local maximum, and if all the minors are negative, the local stationary point is a minimum. (1) I Q.10. Solve the following NLPP: Maximize Z = 4><, + 6x 1 2X 1 2 2 (2) Subjectto' Xl + = 2, Xl' Solution The objective function as well as the constraint are differentiablew.r.t. Xl (3) The constraint can be replaced by another differentiable function such as,
The Lagrangean function can be written as \
L(Xl,X2,A)= 4x l +Xl- 2x f -2XIX2-2xi -A(xl +2x2-2) The necessary conditions for a maxima or minima of the objective function are dL -=0, dXl dL
. dX2 . dL , -=0, and dl.. or 4 4Xl - - 'A = 0, , ... (i) 6 -2Xl - - 2'A = 0, ... (ii) and - (Xl + - 2), = 0 , ... (iii) . 80-2.4 65 80-2.4 Solving (i) and (ii), we get x 1 = 1/3 which when substituted in (iii) gives = 5/6 and for this solution A = 1. The sufficient conditions for determining whether this solution resu1ts in maximizationor minimization of the objective function involve the solution of (n - 1) =2 -1 =1principal minor of thefollowing determinant of order 3 : o 1 2 = 1 -4 -2 =-1(-4+4)+2(-2+8)=12 2 -2 -4 Since D3 is positive, the solution x 1 = 1/3 and =,5/6 maximizes the objective function. 6X! -2(jr = 150 = 4.166. 36 Q.11 Solvethefollowing N.LP.P. Subjectto x 1 + =4, Solution. The Lagrangean function for the above problem can beformed as L (X, A) = 5Xl +x2 -(XI-X2)2'-A(Xl +x2 -4) The necessary conditions for the objective function to be maximum or minimum are ... (i) ... (ii) and .. , (iii) BC-2.4 dL -=-(Xl +x2 -4)=0 a').. 66 in -1) tive num xl=-,x2=- andA.=3 2 2 o 1 .. Now D.3 =11 -2 =-1(-2-2)+1(2+2)=4+4=8 1 2 Since D3 is positive, the solution Xl = 5/2 and = 3/2 maximizes the objective function and z.... =25 +1.-1=13 G 2 2 2 2 2 2 Solve the following N.L.P.P. by ushlg Lagrangean multipliers method: 2 2 2 Minimize Z =Xl +X3' Subjectto 4x l 14
Solution The Lagrangean function for the given problem can bewritten as ':. 2 2 2 '1 2 L(Xt,X2,x3,1\,)=xl +x2 +x3 -I\, (4xl +x2 +2x3 -14) The necessary conditions the above convex function to be minimum are dL = 2x l -4')..,=0, ... (i) dx} dL = 2x 2 2')..,X2 = 0, ... (ii) dX2 dL ... (iii) dX3 =2x3 -21...=0, 67 BC-2.4 6 The solution ofthese equations yields 5 . 3 -2.4 ... (iv) 'A=Xl and from (iii) iI, = From (i) 2 From (ii) (1 - iI,) =0 i.e., = Ooril, = 1. When From(iv) 4 (2 iI,) +0 + 2 iI,= 14 or iI, = 14. This gives the solution X, = 2.8, 0, X3 = 1.4, Zmin. =9.8 with iI, =1, x when 1 =2, from (i) from (iii) x 1 =2 andfrom (iv) 8 + Xz2 +2 =14 orXz2 = 4 or x 1 =2 since 0, Xz = -2 is discarded. Thesecond solution is Xl =2, Xz =2, = 1, giving 2.mn. = 9. Since (Xl' Xz, =(2, 2, 1) givesthesmaller value of Zmin. It is the optimal solution. Q.13 Obtain the necessary and sufficient conditions for the optimal solution of the following problem. What isthe optimal solution? Minimize Z =2e 3x + l + e'b2+3, Subject tothe constraint x 1
X"Xz20 BC-2.4 68 Solution Ition. )f the The objective function as well as the constraint are differentiable with respect to x 1 and X:! and the Lagrangean function for the above problem can be formed as " L(X,A)=2e 3 ;q+1 +e 2X2 + 3 -A(xl+x2 -5) The necessary and sufficient conditions for maximization or minimization of Z = f (X1' ><2) can be obtained as aL =6 e 3 x:t+ 1 -A=O. . .. (i) ax} aL =2e2x2+3 -A=O, ... (ii) aX2 aL and dA =-(xl +x2 -sy=O: ... (iii) From (i) and (ii), 6;Xl+l =2e 2 .'t2+ 3 and from (iii) X, + X:! = 5. Xl + X:! = 5. 3e3xt +1 =e13-2x1 or loge3+3xl +1=11.....2xl or 1 xl =- (12 .,-log e 3), or 5 . 1 1 x2 =5--(12-1oge 3) =-(13+loge 3). and 5 5 o 1 1 Now ~ =11 18e 3 XJ. +1 o 1 O.4e 2x \+3 =-: 1(4e 2 ;q +3)+1(-18e 3x , +1) . ." B0-2.4 69 C-2.4 =-ve. [Sincethe expression within parenthesis is positive for all values ofX1 andx 2 ] V Q.1: Solve the following NLPP using the conditions: . 22 Maximize Z = 2xI -7x2 +12xlx2, Subject to 2x, + s 98, Solu x" Solution . The given problem can be put as f(X)=2xr -7.q +12xlx2' h(X)=2xl +5X2 -98. The Kuhn-Tucker conditions are df _ Iv dh =0 dxl (hI ' . CJf _ Iv .!!!..-=o, dx2 dX2 . Ivh(X) = 0, heX) ::; 0, . :0 'f71 Applying these conditions, we get 4x, + ... (i) = 0, ... (ii) A (2x, + =0, ... (iii) 2x, + ... (iv) ... (v)
. From equation (iii) .either A =0 or 2x, + - 98 = 0 BC-2.4 70 0.15. Solution
When A = O,equations (i) and (ii) give X 1 =X 2 =0 which does not satisfy condition (iv). Thus a feasible solution cannot be obtained for A = O. When this equation along with (i) and (ii) gives the solution, and = 2 with A = 1 00 and 2...sx; = 4,900. Maximize Z= lOx! +4X2-2xf-xi, Subject to 2xl
2 2 We have, f(X)=lOxl +4x2 -2Xl -X2, h(X)=2xl +x2 -S. The Kuhn-Tucker conditions for the maximization problem are df(X) _ AiJh(X) =0 ax ' J J Ah(X) =0,
Applying these conditions, we get ... (i) 10-4xl-2A=0, ... (ii} 4-2x2 -A=O, A(2xl +x2 -S)=O, ... (iii) ... (iv) 2Xl +x2 ... (v) AS-;O. From (iii) either A=0 or 2xl +x2 - 5=O. When A=O, the solution of (i) and (ii) gives Xl = 2.5 and = 2.0 which does not satisfy the equation (iv). Hence A= 0 does not yield a feasible solution. 71 BC-2.4 11 4 4 When 2x 1 +X:!-5=0 and 1..=0, the solution of (i), (ii) and (iii) yields, XI ='6,.\2 ==3,A,==3' which satisfies all the necessary conditions. It can be verified that the objective function is concave in X, while the constraint is convex in X. Thus these necessary conditions are also the sufficient conditions of maximization of f(X). Solution . .. 11 4 .. 91 The optimal solution IS xl =6,x2 =3"' whlchglvesZ max . = - Optimize Z = 2X1 + 3X:! (X12 + X:! 2 + X3 2 ) , - Subject to We have . 2 2 2 f(X)= 2xl +3x2-Xl -x2 -x3, h1(X)=Xl +x2 -1, and h 2 (X)= 2xl +3x2 -6. Before applying the Kuhn-Tucker conditions, it is essential to determine whether the problem is of maximization or of minimization type. For that, we construct the bordered Hessian matrix. H B ~ ( O pT ~ L+n)(m+nl 0 0 1 0 0 0 2 3 0 n=3, =-10. m=2, = 2 -2 0 0 ll-m=l, 1 3 0 -2 0 2m+l=5. 0 0 0 0 -2 For maximization, the sign of HB should be (-1 }m+1 i.e., -ve, while for minimization it So should be (-1t Le., +ve. Since HB < 0, the solution point should maximize the objective function. 72 BC-2.4
3' 3' traint is tions of IIJhether ructthe zation it lize the BC-2.4 The Kuhn-Tucker conditions forthis case are m fj(X)- LAi h} (X)=O, i=l
Therefore, we have ... (i) 2-2xI-Al-2A2 =0, 2-2x2 -At -3"-2 =0, ... (ii) ... (iii) - 2X3=0, ... (iv) Al (Xl +x2 -1)=0, "-2 (2xt +3x2 -6)=0, ... (v) Xl ... (vi) 2x} +3x2 ... (vii) A}>A220. ... (viii) Four solutions corresponding to the following values of Al and A2 can beobtained: (i) A1 =0, "'-2 = 0, (ii) A1 =0, (iii) Al *0, (iv) Al *0, Solution 1 : Al =0, A2 =0. Equations (i), (ii) and (iii) give Xl = 1, x 2 == 3/2, X3 == O.This solution does not satisfy equation (vi) and is, therefore, discarded. 73 BC-2.4 Solution 2: ""I =0,"-2 :;CO. The solution of equations (i), (ii), (iii) and (v) gives l2 18 1 xl =13,X2 =13'X3 O , ~ =13' This solution, again, does not satisfy equation (vi) and is, therefore, deiscarded, Solution 3: ""I :;cO, "-2 =0. I ' ! The solution of equations (i), (ii), (iii) and (iv) yields 1 3 3 xl ='4,X2 ='4,X3 =0, A.I =2" . This solution satisfies all the conditions, and gives 17 Z= . 8 Solution 4 : ""} :;cO,"-2:;cO The solution of equation (i), (ii), (iii), (iv) and (v) gives xl =-3'X2 =4'X3 =0'''"1 =-34,A,z =13. This solution violates conditions (viii) and is thus infeasible and is, therefore, discarded. . Since only one solution satisfies all the conditions, the same is the optimal solution. 1 . 3 xl =-,x2 =-,x3 =0 4 4 I , . , 17 and Zmax = 8 a.17. Use the Kuhn-Tucker conditions to solve the following non-linear programming problem: Maximize Z= 2xI +3x2 6 , Subject to the constraints 2x} +3x2 ~ 6 , 74 BC-2.4 Irded. :herefore, solution. Iramming BC-2.4 Solution 2 We have f(X)=::2xt-xt +x2, h t (X)=2xl-3x2 -6, and h2(X)=2xl +x2 -4. The Kuhn-Tucker conditions for a maximization problem are m fj(X)- L ~ h . ~ (X)=O. i=1 'A;hi(X)=O. hi (X):<:;;O, Ai $0. Applying these conditions to the given functions, we get 2-2xl-At -2"-2 =0, ... (i) 1-3A,. -"'z =0, ...(ii) ... (iii) Al (2xl +3x2 -6)=0, ...(iv) "'z(2Xl +x2 -4);:::0, . .. (v) 2XI +3x2 -6$0, ...(vi) 2xl +x2 -4$0, AI,"'z O . . .. (vii) The following situations can nowarise : Case 1 : Al =0, "-2 =0, Case 2 : Al =0, A2:;t:0. Case 3: At :;to, A2 =0, Case 4 : Al :;to, A2 :;to, Case 1 : When A1 =0, A 2 =0, condition (ii) is violated and this solution is, therefore, discarded. Case 2: When A1 =0, A2* 0, solution of equations (i), (ii), (iii) and (iv) gives 75 BC-2.4 Xl = 0, Al =0,1.. 2 *0,X2 = 4 = 1. As this solution does not satisfy equation (v), it is discarded, Case 3: When 1.,1 *0, 1.,2 =0, solution of equations (i), (ii), (iii) and (iv) yields 2 14 1 xl =3,X2 =9,1..1=3 This solution satisfies all the constraint equations and is thus feasible solution, giving Z = 22/9. Case 4: When A,1*Oand 1.,2*0, the solution of equations (i), (ii), (iii), and (iv) gives 335 xl ="2,X2 =1,1- 1 =4 and '-2 ='-4 This solution violates condition (vii), which is Oand hence is infeasible. Thus the optimal solution is 2 14 .. 22 xl =3,X2 =9' giving l.nax = '9 ( ( ( ( ( ( ( ( ( ( BC-2.4 76 SECTION C SHORT ANSWER TYPE QUESTIONS 1. What do you mean by a Beam? Ans: A beam' is a structural element that is capable of withstanding load primarily by resisting bending. Generally a beam is of moderate size and is made up of one piece. If the size is large and the beam is made up of many parts joined together, then it is called girder. 2. What are the different types of beams depending upon the end conditions? Ans: Depending upon the end conditions the various types of beams are: i) Cantilever Beam: A beam which is fixed at one end and free at the other end. . ii) Simply Supported Beam: Abeam supported or resting freely on the supports at its both ends, is known as simply supported or freely supported beam. The supports may be either in the form of columns or walls. iii) Overhanging Beams: A beam in which its end portion is extended beyond the support, is known as overhanging beam. A beam may overhang on one side or on both sides. iv) Fixed Beams: Abeam whose both ends are fixed or built in walls, is known as fixed beam or encastred beam. v) Continuous Beam: A beam supported on more than two supports is known as continuous beam. The extreme left and right supports are called end supports and all the remaining are known as intermediate supports. 3. What is loading diagram? Ans: A loading diagram shows all loads applied to the beam which includes the service loads as well as the reaction loads. The service loads are loads put on the building during its use. These include dead, live, roof live, snow, wind, earthquake, and other types of loads. 4. Define Shear Force and Bending Moment. Ans: Shear force at a section in a beam is the normal force that is trying to shear off the section and is obtained as the algebraic sum of the forces acting normally to the axis of the beam to the right or left ofthe section. Bending moment at a section in a beam is the moment that is trying to bend the beam and is obtained as the algebraic sum of the moments of all forces acting either to the right or left of the section. 5. What do you mean by statically determinate structure? Ans: In statics, a structure is statically indeterminate when the static equilibrium equations are not sufficient for determining the internal forces and reactions on that structure. Based on Newton's laws of motion, the equilibrium equations available for a two-dimensional body are 8C-2.4 136 a) the vectorial sum of the forces acting on the body equals zero. This translates to r. H =0: the sum ofthe horizontal components ofthe forces equals zero; r. V =0: the sum of the vertical components offorces equals zero; b) the sum ofthe moments (aboutan arbitrary point) of all forces equals zero. 6. Discuss types of Loading on a Beam. Ans: Loads applied to the beam may consist of a concentrated load (load applied at a point), uniform load, uniformly varying load, or an applied couple or moment. These loads are shown in the following figures. CoocenbGlBd l.DaJs uniform load ... M .
Unifoonly Vatying iJl;3d' - AppiJed Couple 7. What is Principle of Superposition? Ans: It states that when a number of forces (loads) are acting on a body, then the resulting strain will be equal to algebraic sum of strains caused by individual loads. While using this principle for an elastic body which is subjected to a number of tensile or compressive forces at different sections along the length of the body, first the free body diagram of individual section is drawn. Then the deformation of each section is calculated. The total deformation of the body will be then equal to the algebraic sum of deformationsofthe individual sections. 8. Explain First theorem ofCastigiiano. Ans: In any beam or truss subjected to any load system, the deflecti()n at any point r is given by the partial differential coefficient of the total strain energy stored with respect to a force Pr, acting at the point r in the direction in which the deflection is desired. Albert Castigliano an Engineer ofItalian Railways offered his theorems in 1876. 9. What do you mean by Maxwell's Reciprocal Theorem? Ans: In any-beam or truss the deflection at any point D due to a load W at any other point C is the same as the deflection at C due to the same load W applied at D 137 BC-2.4 138 BC-2.4 James Clerk Maxwell, a Cavendish Professor at Cambridge University offered this lawin 1864. 10. Define Elasticity and Elastic Limit. Ans: A material is said to be perfectly elastic if the deformation produced in it by external forces, completely disappears on the removal of external forces. This property by virtue of which certain materials return back to their original position after the removal of the external forces, is called elasticity. SECTION D LONG ANSWER TYPE QUESTIONS 01 : Design Uniform column of tubular section (Fig.) to carry a compressive load P = 2500 kgf for minimum cost. The column is made up of a material that has a yield stress (cry) of 500 kgf/cm 2 , modulus of elasticity (E) of 0.85 x 10 6 kgf/cm 2 , and density (p) of 0.0025 kgf/cm 3 The length of the column is 250 cm. The stress induced in the column should be less than the buckling stress as well as the yield stress. The mean diameter ofthe column is restricted to lie between 2 and 14 cm, and columns with thicknesses outside the range 0.2 to 0.8 cm are not available in the market. The cost of the column includes material and construction costs and can be taken as 5W + 2d, where W is the weight in kilograms force andd is the mean diameter of the column in centimeters. Solution: The design variables are the mean diameter (d) and tube thickness (t) : ... (E1) X= { : ~ H : } The objective function to be minimized is given by ... (E2) f{X) = 5W + 2d = 5p17tdt +2d =9.82x}x2 +2x} I J A r Section A A p Fig. Tubular column under compression. The behavior constraints can be expressed as stress induced ~ yield stress stress induced buckling stress The induced stress is given by P 2500 Induced stress = cri -=-- ... (E3) . ndt nx}x2 139 BC-2.4 The buckling stress for a pin-connected column is given by Enter buckling load Tt 2EI . 1 Buckling stress = .0' b - - ... (E4) - cross-sectionai area T -Tt dt Where I =second moment of area ofthe cross-section ofthe column. Tt 4 4 = -(do -d) 64 ,I Tt 2 2 = 64 (do -d i )(d o +di)(d o -:d i ) = 2:..[(d +t)2 +(d -t)2].[(d +t)+(d -t)][(d+t)-(d'-t)] 64 . . .. (ES) = The Thus the behavior constraints can be restated as 2500 \' gl (X)=---500:::;O ... (E6) TtXlx2 2 . 6 2 2 g2 (X)= 2500 _ Tt (0.85 x 10 )(xl +x2) <0 ... (E7) TtXlx2 8(250)2 The side constraints are given by 2:::;d:::;14 which can be expressed in standard form as ... (E8) . g3 (X)=-xl =2.0:::;0 g4 (X)=-xl-14.0:::;0 ... (E9) gs (X)=-x2 -0.2:::;0 ... (E10) g6 (X)=x2 -0.8:::;0 .. .(E11) Since there are only two design variables, the problem can be solved graphically as . shown below. First, the constraint surfaces are to be plotted in a two;dirnensional design space ' where the two axes represent the two design variables 1 2500 gl (X) = --- 500:::; 0 TtXlx2 140 BC-2.4 that is, , xlx2 ;::::,1.593 Thus the x, x 2 = 1.593 represents the constraint surface gl (X) =O. This curve can be plotted by finding several points on the curve. The points on the curve can be found by giving a series of values to X, and finding the corresponding values of X 2 that satisfy the relation X, X 2 = 1.595 :
2.0 4.0 6.0 I 8.0 10.0 12.0 14.0 . Xl x2 0.7965 0.3983 0.2655 I 0.1990 0.1593 0.1328 0.1140 These points are plotted and a curve P,Q, passing through all these points is drawn as shown in Fig. and the infeasible region, represented by g, (xO or X, X 2 <1.594, is shown by hatched lines.* Similarly, the second constraint g2 (X):$;O can be expressed as X, X 2 (x/ + X22);;:: 47.3 and lying on the constraint surface gz(X) =0 can be obtained as follow [for xlx2 (xt +x!)=47.3] - XI 2 4 6 8 10 12 14 x2 2.41 0.716 0.219 0.0926 0.0476 0.0274 0.0172 --- These points are plotted as curve P,Q2' the feasible region is identified, and the infeasible region is shown by hatched lines as shown in Fig. The plotting of side constraints is very simple since they represent straight lines. After plotting all the six constraints, the feasible region can be seen to be given by the bounded area ABCDEA. * The infeasible region can be identified by testing whether the origin lies in the feasible region. Yield constraint d(x2 l gl (x) = 0 0.91 0.8 0.7 0.6 0.5 BllCkli ng constraint t g2(X}=0 jl(X)=O E -.,' 0.4 Optimum point 'as (5.44. 0.293) 0.3 0.2 Ice 0.1 0 i -- , .. :;r "d(>q) 2 4 6 8 10 14 Fig. Graphical optimization 4 141 BC-2.4 Next, the contours of the objective function are to be plotted before finding the optimum point. Forthis, we plotthe curves given by f(X) =9.82, X i X 2 + 2Xi =C =constant for a series of values of c. By giving different values to c, the contours of f can be plotted with the help ofthe following points. Forf(X) 9.82, X i X 2 + 2Xi =50.0 I Xl 0.1 0.2 0.3 0.4 0.5 0,6 0,7 , 0.8 ! ~ I x2 16,77 12,62 10.10 8.44 I 7.24 I 6.33 I 5.64 i 5.07 I I Forf(X) 9.82, X i X 2 + Xi = 40 0.1 0.2 OJ 0.4 0.5 0.6 0.7 ~ . 400 10.10 8.08 6,75 5.79 5.0 4.51 L _ ~ . Forf(X) 9.82, X i X 2 + 2Xi = 31.58 (passing through the corner point C) : I Xl 0.1 0.2 OJ 0.4 0.5 0.6 I 0.7 I 0.8 \ x2 10.57 I 7.96 6.38 5.33 4.57 4.00 3.56 l 3.20 I I I I \ Forf(X) 9.82, X i X 2 + 2Xi =26.53 (passing through the corner point B): 8.88 6.69 I 5.3 I 4.48 3.84 3.36 I I 2.99 I 2.69 , I I I XI I 0.1 0.2 I 0.3 0.4 0.5 I 0.6 I 0.7 I 0.8 I i x2 . ' ~ , XI 0.1 0.2 ! 0.3 0.4 0.5 0.6 0.7 I 0.8 I x2 6.70 5.05 4.04 I 3.38 2.90 2.53 2.26 2.02 i I Forf(X) 9.82, X i X 2 + 2Xi = 20.0 These contours are shown in Fig. and it can be seen that the objective function cannot be reduced below a value of 26.53 (corresponding to point B) without violating some of the constraints. Thus the optimum solution is given by point B with d* =Xi =5.44 cm and f =x 2 =0.293 cm with f min=26.53. 142 8C-2.4 1 A rocket is designed to travel a distance of 12s ina vertically upward direction. The t thrust of the rocket can be changed only at the discrete points located at distances of 0, s, 2 s, 3 s, ... , 11 s. If the maximum thrust that can be developed at point I either in the positive or negative direction is restricted to a value of FP formulate the problem of minimizing the total time oftravel under the following assumptions: 1. The rocket travels againstthe gravitational force. 2. The mass ofthe rocket reduces in proportion to the distance traveled. 3. The air resistance is proportional to the velocity ofthe rocket. Solution: Let points (or control points) on the path at which the thrusts of the rocket are changed be numbered as 1, 2, 3, ... , 13 (Fig. ). Denoting XI as thrust, VI the velocity, a, the acceleration, 13.125 12. lis II lOS 10.9s /
3 ., 2s 2 .5 1 .0 ?, Control Distance tTom points starting point Fig. Control points in the path of the rocket. And nil the mass of the rocket at point i, Newton's second law of motion can be applied as net force on the rocket =mass xacceleration This can be written as Thrust - gravitational force - air resistance =mass x acceleration Or x -mg-k1v =ma I I I I I ... {E1) Where the mass mj can be expressed as . m=m l-k 2 s 1 1 ... {E2) 143 BC-2.4 and kJ and k2 are constants. Equation (EJ can be used to express the acceleratin, a i ,as 1 klVi a . ... (E3) 1 m. m. ' I 1 If t, denotesthe time takenby the rocket to travel from point i to point i + '1, the distance traveled between the point I and i + 1can be expressed as ... (E4) or , from which ti can be determined as 2--( -vi Vi +2S1 g_._l) \111- m ... (E5)
t.= I I I Xi k 1 Vi "',-g-,'" mi n1i Of the two values given by Eq. (E5), the positive value has to be chosen for Ii . The velocity of the rocket at point i + 1, v j +1, can be expressed in terms of Vi as (by assuming the acceleration between points I and i + 1 to be constant for simplicity) V' 1=v +a-t ... (E6) l+ ' I I I The substitution ofEqs. (E3) and (Es) into Eq. (E6) leads to , 2 ( Xi k1vi ') ... (E7) Vi+l = Vi +25' --g--j mj From an analysis of the problem, the control variables can be identified as the thrusts and the state variables as the velocities, Vi . Since the rocket starts at point 1 and stops at point 13, ... (E8) Thus the problem can be stated as an optimization problem as Find X = which minimizes 144 BC2.4 a It subject to 12 . . 12!-Vi+ IV1.+.2s(Xi _g_k1Vi) f(X) =Lti=L mi 1ni i=1 i=1 ( Xj k 1 v. ) _._g __1 l mi mi mi+l =mi -k 2 s, ;=1,2, ... ,12 . I 2 ( Xi kIV; ) . vi +2s .--g:--. ,1=1,2, ... ,12 mi mi Ixi j:::;P i , i=1,2, ...,12 VI =v13 =0. Q. 3: The step-cone pulley shown in Fig. is to be designed for transmitting a power of the least 0.75 hp. The speed of the input shaft is 350 rpm and the output speed requirements are 750, 450, 250, and 150 rpm for a fixed center distance of a between the input and output shafts. The tension the tight side of the belt is to be kept more than twice that on the slack side. The thickness pf the be is t and the coefficient of friction between the belt and the pulleys is 11. Formulate the problem of finding the width and diameters of the steps for minimum weight. Solution: The design vector can be taken as d l d 2 X= {d 3 d 4 w .. Where d l is. the diameter of the ith step on the output pulley and w is the width of the belt and the step.
"I = 350 Fig. Step-cone pulley BC"2.4 The objective function is the weight of the step of the step-cone pulley system: +d 2 +d 2 +d,2 +d,2 +d ,2 +d ,2 ) . 4 1 2 3 4 1 .2 3 4
+ [1 + d3 [1 + ...(E,) where p is the density of the pulleys and d,' is the diameter of the it step on the input pulley. To have the belt equally tight on each pair of opposite steps, the total length of the belt must be kept constant, for all the output speeds: This can be ensured by satisfying thefollowing equqlity constraints: . C1-C2=O ... (E2) C1-C3=O ... (E3) C1-C4=O ...(E4) Where C i denotes length of the belt needed to obtain output speed N, (i =1 , 2, 3, 4) and is given by (Ni )2 2 --1 d nd ( N.) N I 1 Ci=_l ll+- + +2a 2 N 4a where N is the speed of the input shaft and a is the center distance between the shifts. The ratio oftensions in the belt can be expressed as =el!8i
where T\and Ti2 are the tensions on the right and slack sides of the ith step, ).l the coefficient of friction, and {Ii the angle of lap of the belt over the ith pulley step. The angle of lap is given by 146 BC-2.4 )ut the by 3,4) m the I! the p.The 3C-2.4 and hencethe constraint onthe ratio oftensions becomes 1 1 { IN) d. I]} i = 1,2,3,4. . .. (Es) exp {[ fC-2sin- l N - 1
The limitation on the maximum tension can be expressed as T{ i = 1,2,3,4 ... (Es) where s is the maximum allowable stress in the belt and t is the thickness of the belt. The constraint on the power transmitted can be stated as (using fb, for force and ftfor linear dimensions) (T{ - 33,000 . which can be rewritten using T{ ::;::; stw (upper bound used for simplicity) and Eq. . . .... . " . ( . ( . {liN' ') do}']" . o( 350\. . stwll-exp -J.tlfC-2sll1 _1_1 )'. N . 2a . 33,000/ . ,:' " ... (Ei) Finally, the lower bounds on the variables can be taken as
... (Ea) d i
i =1,2,3,4 ... (Eg) As the objective function, (E1)' and most of the constraints, (E 2 ) to (E9)' are nonlinear functions of the design variables d,. d2f d 4 and w, this problem is a nonlinear programming problem. Geometric Programming Problem Definition : A function h (X) is called a posynomial if h can be expressed as the sum of power terms each ofthe form. ait ai2. ain CiXl X2 "'Xn where c j and all are constants with c j > 0 and XI> O. Thus a posynomial with N terms can be expressed as xU X12 "In aNt llN2 aNn h(X) = CtXl x2 ",xn +",+cNx l X2 ",XlI 147 BC-2.4 A geometric programming (GNP) problem is one in which the objective function and constraints are expressed as posynomials in X. Thus GMP problem can be posed as follows Find X which minimizes (1) where No and Nk denote the number of posynomial terms in the objective and kth constraintfunction, respectively. Q.4: Four identical helical springs are used to support a milling machine weighting 5000 lb. Formulate the problem of finding the wire diameter (d), coil diameter (D), and the number of turns (N) of each spring (Fig.) for minimum weight by limiting the deflection to 0.1 in and the shear stress to 10,000 psi in the spring. In addition, the natural frequency of vibration of the spring is to be greater than 100Hz. The stiffness of the spring (k), the shear stress in the spring (r), and the natural frequency of vibration ofthe spring (f,) are given by i l F I+- 1)-1 Fig. Helical Spring. 8FD 't=K s - nd 3 148 BC-2.4 l
I, e iI where G is the shear modulus, F the compressive load on the spring, wthe weight of the spring, p the weight density of the spring, and Ks the sher stress correction factor. Assume that the materialis spring steel with G =12 x 10 6 psi and p =O.S Ib/in 3 , and the shear stress correction factor is Ks =1.05. Solution: The design vector is given by: Xl} .{d} X= {
And the objective function by nd 2 f(X) =weight =4nDNp ... (E 1 ) The constraint is can be expressed as . F 8FD 3 N Deflection =- 4 <0.1 k dG That is, . d 4G gl (X): 3 >1 ... (E3) 80FD N \ .JGi d natural frequency = r;;-:::: -2- 100 2v 2 pn D N that is, g3 (X): .JGid >1 ... (E4)
Since the equality sign is not included (along with the inequality symbol, in the constraints of Eqs. (EJ to (E4) , the design variables are to be restricted to positive values as ., d > 0, D > 0, N >0 ... (Es) By substituting the know data, F= weight ofthe milling machine /4 = 1250 Ib, P =O.Slb/inS, G = 12x10 6 psi,and Ks = 1.05, Eqs. (E 1 ), to (E4) become (0.3)1 2 DN 0.7402xfx2X3 ... (E6) 4 d202x106) 120 4 -3 -1 1 g (X) = == xx x> ... (E7) I 80(1250)D 3 N 1 2 3 ,xi' > 1 ... (E8) g2 (X) D 149 BC-2.4 4 , : ... (E9) F >1 200 2pnD 2 N . It can be seen thatthe objective function, f (X), and the constraintfunctions, 91 (X) to g3(X), are posynomials and hence the problem is a GMP problem. Quadratic Programming Problem. A quadratic programming problem is a nonlinear programming problem with a quadratic objective function and linear constraints. It is usually formulated as follows: n n n F(X) = c-Lqlxl +I,LQijXiXj i=l i=l i=l n subjectto "ax=b . L.JlJ J"-12 i=l J - , , .. , m Xi c:: 0, i=1,2, ... ,n where c, qb ql' ail and b l are constants. Q5: A manufacturing firm produces two products. A and B, using two limited resources. The maximum amQunts of resources 1 and 2 available per day are 1000 and 250 units, respectively. The production of 1 unit of product A requires 1 unit of resource 1 and 0.2 unitof resource 2, and the production of1 unitofproduct B requires 0.5 unit of resource 1 and 0.5 unit of resource 2. The unit costs of resources 1 and 2 are given bytherelations (0.375 - 0.00005u 1 ) and (0.75-0.00001 u 2 ), respectively, where u1 denotesthe number of units of resource i used (i = 1,2). The selling prices per unit of products A and B, PA and PB, are given by PA = 2100 0.0005xA 0.00015xB PB = 3.50-0.0002xA -0.0015xB.. .. . ' Where X A and xBindicate, respectively, the number of units of products A and B sold. Formulate the problem of maximizing the profit assuming that the firm can sell all the units it manufactures Solution: Let the design variables bethe number of units of products A and B manufactured per day: The requirement of resource 1 per day is (XA + 0.5 XB ) and that of resource 2 is (0.2XA +0.5 xB ) and the constraints on the resources are o a ir
ire ) 1 Jct of :ei ue j B
red 2XA
XA ... (E1) O.2xA + O.5xB 250 ... (E2) The lower bounds on the design variables can be taken as
... (E3)
...(E4) The total cost of resources 1 and 2 per dayis (xA +0.5xB)[0.375-0.00005(xA +0.5xB)] +(0.2xA +0.5xB)[0.750-0.0001(0.2xA +0.5xB)) and the return per day from the sale of productsAand B is xA (2.00-0.0005xA -0.00015xB)+xA (3.50-0.0002xA -0.0015xB) The total profit is given by the total return minus the total cost. Since the objective function to be minimized is the negative of the profit per day,f(AJ is given by f(X)=(xA + 0.5xB)[0.375-0.00005(xA+0.5xB)) +(0.2xA + 0.5xB)[0.750-0.0001(0.2xA +0.5xB)) -xA(2.00- 0;0005xA ..L0.000l5xB) -xB(3.50-0.0002xA -0.0015xB) ...(E5) As the objective function [Eq. (Es)] is a quadratic and the constraints [Eqs. (E 1 ) to (E4) are linear, the problem is a quadratic programming problem. 06: A Scaffolding system consists of three beams and six ropes as shown in Fig. Each of the top ropes A and B can carry a load of WI' each of the middle ropes C and D can carry a load ofW l , and each ofthe bottom ropes E and F can carry a load ofW3. Ifthe load acting on beams 1,2, and 3 arexI,x l and X 3 r.espectively, as shown in Fig. formula the problem of finding the maximum load (Xi +Xl +xJ that can be supported by the system. Assume that the weights ofthe beams 1,2, and 3 are WI> W l , and W 3 ; respectively, and the weights of the ropes are negligible. Solution: Assuming that the weights of the beams act through their respective middle points, the equations of equilibrium for vertical forces and moments for each of the three beams can be written as : For beam 3: TE+TF =x3 +w3 x3(31) +w3 (21)- TF (4/)=0 151 BC-2.4
1-21 2i-1 '--_----.___----.___---,-__ Fig. Scaffolding system with three beams. For beam 2: X2 (/)+ w2 (I) +TE (I) - T D (2/)=0 For beam 1: T A +TB - TD - TF =xl +WI xl (3/)+ wl - TB (91)+ Tc (21)+ T D (4/)+ TF (7/)=0 where Ti denotes the tension in rope i. ;r'he solution of these equations gives 3 1 T F =-x3+- w3 4 2 1 1 TE =-x3 +-'"':3 4 2 1 1 1 1 TD =-x2 +-x3 +-w2 +-x3 2 8 2 4 1 1 1 1 Tc =-x2 +-x3 +-).12 +-x3 2 8 2 4 1 1 2 1 1 5 TB=-xl +-x2 +-x3 +-wl +-w2 +-w3 3 3 3' 2 3 9 2 2 1 1 2 4 TA =-xl +-x2 +-x3' +-wl +-w? +,,-w3 3332 The optimization problem can be formulated by choosing the design vector as s 152 BC-2.4 Since the objective is to maximize the total load fiX)= -(xl +X2+ X 3) ... (E1) The constraints on the forces in the ropes can be stated as T A o;W 1 ... (E2)
... (E3) T C o;W 2 ... (E4) T D o;W 2 ... (E5)
... (E6) T F 5W 3 ... (E7) Finally, the nonnegativity requirement ofthe design variables can be expressed as xl ;:::0 ... (E8) X2 ;:::0 X3 ;:::0 Q7: A cargo load is to be prepared from five types of articles. The weightwi. volume vi and monetary value ci of different articles are given below. -
Article Type Wj Vi Ci 1 4 9 5 2 8 7 6 3 2 4 3 4 5 3 2 5 3 8 8 Formulate the problem find the number of articles xi selected from the ith type (I = 1. 2. 3.4, 5). so that the total monetary value of the cargo load is a maximum. The total weight and volume of the cargo cannot exceed the limits of 2000 and 2500 units. respectively. Solution: Let xi be the number of articles of type i (i =1 to 5) selected. Since it is not possible to load a fraction of an article, the variable xi can take only integer values. 153 BC-2.4 II The objective function to be maximized is given by . f(X) = 5XI +6x2 +3x3 +2X4 +8X5 ... (E1) and the constraints are ... (E2) ... (E3) XI;:::O and integer, i= 1, 2, .... 5 ... (E4) Since xi constraints to be integers, the problem is an integer programmer problem. Q8: Formulate the problem of designing a minimum-cost rectangular under-reinforced . concrete beam that can carry a bending moment M with a probability of at least 0.95. The costs of concrete, steel, and formwork are given by C. = $ 200/m 3 , C s = $5000/m 3 and C f =$40/m 2 of surface area. The bending moment M is a probabilistic quantity and vary between 1X10 5 and 2X10 5 N-m with a uniform probability. The strength of concrete and steel are also uniformly distributed probabilistic quantities whose lower and upper limits are given by f,= 25 and 35 MPa Ix= 500 and 550 MPa \ Assume that the area of the reinforcing steel and the cross-sectional dimensions of the beam are deterministic quantities. Solution: The breadth b in meters, the depth d in meters, and the area of reinforcing steel A, in square metres are taken as the design variables x 1 , ~ and X3 respectively (Fig.). The cost ofthe beam per metre length is given by fiX) =cost of steel + cost of concrete + cost of formwqrk .. ( E 1 ) As I d / l ~
1 I.. b III Fig. Cross-section of a reinforced concrete beam. 154 BC-2.4 c )f
.). 4 The resisting moment of the beam section is given by M =A f, (' d_O.59 As }S) B s. S (sb and the constraint on the bending moment can be expressed as [1.70] P[MB -M <:Ol:P[ Asfs(d-O.59 jj J-M;'O];'0.95 ...(E2) where P [ ... ] indicates the probability of occurrence of the event [ ... ]. To ensure that the beam remains under-reinforced the area of steel is bounded by the balanced steel areaA,,(b 1 as A ' < A(b) s s A(b) =(0.542) Ie bd 600 s 600+is where Since the design variables cannot be negative, we have d"?:.O
... (E3) ... (E4) Since the quantities M/c, and1. are nondeterministic, the problem is a stochastic programming problem. Q. 9 : A retail store stocks and sells three different models of TV sets. The store cannot afford to have an inventory worth more than $45,000 at any time. The TV sets are ordered in lots. It costs $ aJ for the store whenever a lot of TV model is ordered. The , cost of one TV set of model j is Cj The, demand rate of TV model j is d j units per year. The rate at which the inventory costs accumulate is known to be proportional to the investment in inventory at any time, with qJ = 0.5, dehoting the constant of proportionality for TVG modelj. Each TV set occupies an area of S J = 0040 m 2 and the maximum storage space available is 90 m 2 The data known from the past experience are given as follows: l TVModelj 2 3 Order cost a, ($) 50 10 100 Unit cost ci ($) -40 120 80 Demand rate, d; 800 400 1200 -------- 155 BC-2.4 FOimulate the problem of minimizing the average annual cost of ordering and storing the TV sets. Solution: Letxj denote the number of TV sets of model j ordered in each lot (j = 1,2,3). Since the demand rate per year of modelj is dj, the number of times the TV model j needs to be ordered is dJ/x j The cost of ordering TV modeljperyear is thus aAlxyj= 1, 2,3. The cost of storing TV sets ofmodelj per year is q)c jX / 2 since the average level of inventory at any time during the year is equal to c jX / 2. Thus the objective function (cost of ordering plus storing) can be expressed as , (aidl qlclxl ') (q2 d 2 Q2 C 2 X 2) (a3d3 Q3 c 3 X 3" f(X)= -+-- + --+-- + -+--1 ... (E1 ), Xl 2 X2 2 x3 2) where the design vector X is given by ... (E2) The constraint on the volume of inventory can ~ stated as clxl +c2x2 +c3x3 ::;45,000 ... (E3) The limitation on the storage area is given by ... (E4) Since the design variables canot be negative, we have x j 20, j=1,2,3 ... (ES) By substituting the known data the optimization problem can be stated as follows: Find X which minimizes (40,000 10 ) (32,000 30 ) e,20,000 20 ) + xl + + x2 + + x3 fiX) ... (E6) Xl x2 ' x3 subject to gl eX) = 40Xl +12x2 +80X3 ::; 45, 000 ... (E7) ... (EB) g2 (X)= OAO (xl +X2 +x3) ::;90 ... (E9) g3(X)=-xl ::;0 ... (E10) g4(X)=-x2::;O ... (E11) gs(X)=-xs :::;0 156 BC-2.4 ~ s a e s: It can be observed that the optimization problem stated in Eqs. (E6) is a separable programming problem. 010: . A uniform column of rectangular cross section is to be constructed for supporting a water tank of mass M (Fig.). It is required (1) to minimize the mass ofthe column for economy, and (2) to maximize the natural frequency of transverse vibration of the system for avoiding possible resonance due to wind. Formulate the problem of designing the column to avoid failure due to direct compression and buckling. Assume the permissible compressive stress to (j . 1 b ~ , ,- Cross-section ofthe colwnn Fig. Water tank on a column . .. Solution: Let xl=b and x 2 =d denote th'ecross-sectional dimensions of the column. The mass ofthe column (m) is given by m == pbdl == plxlX2 ... (E1) where p is the density and I is the height of the column. The natural frequency of transverse vibration of the water tank (ro ), by treating it as a cantilever beam with a tip mass M, can be obtained as [1.68]: . 112 31 ... (E2) 00=1-( 33.) 3 M+ m t 140 . where E is the Young's modulus and I is the area moment of inertia of the column . given by r = = ~ b d 3 ... (E3) 2 157 BC-2.4 .4 The natural frequency of the water tank can be maximized by minimizing- ro . With the help of Eqs. (E 1 ) and (E 3 ), Eq. (E 2 ) can be rewritten as 112 E X 1 X ~ ... (E4) 0) = -----r----=--=----;: 3 4/ ( M + ~ O plxlx2 J The direct compressive stress in the clumn due to the weight of the water tank is given by ... (E5) and the buckling stress for a fixed-free column (cr b ) is given by [1.711 ... (E6) To avoid failure of the column, the direct stress has to be restricted to be less than O"max and the buckling stress has to be constrained to be greater than the direct compressive stress induced. Finally, the design variables have to be constrained to be positive. Thus the multiobjectie optimization problem can be stated as follows: Find which minimizes x - {Xl} - X2 ... (E7) 1/2 ... (E8) subject to ... (E9) 158 BC-2.4 gJ (X)= }Jg _ 1t 2 EXI xlx2 48[2 ...(E10) g3(X)=-Xl ... (611)
...(E12) Q 11: The problem of minimum weight design subject to a constraint on the vertical displacement of nope S (U 1) of the three-bar truss shown in Fig. can be stated as follows: Find X ={Xl} which minimizes x2 . f(X)= p(2J2i)xl +plx2 =80.0445xl-28.3x2 ..(E1) subject to
U max I<t-- I", lOO"---+I+-- I '" 100" 1 1=100" 1 s U2 i P = 0.283 Ib!in 3 E = 30 x t 0 6 psI Umax. = 0.001" i .. PI "'300ib Ut Fig. Three-bar truss. Or 1 ... (E2) g(X) = xl' +.[ix2 159 BC-2.4 Wherep is the weight density, E is Young's modul.us, U max the maximum permissible displacement, X 1 the area of cross-section of bars 1 and 3, x 2 the area of cross section of bar 2, and the vertical displacement of node S is given by 1]/ 1 VI =-- ---:==- ... (E3) E Xl +JiX2 Find the solution of the problem using the optimality criteria method. Solution: The partial derivatives ofland g required by can be computed as of' =80.0445, oj =28.3 aXI . . aX2 og -log -.fi -aX-l = (Xl +.fi X2)2 'OX2 = (Xl +.fiX2)2 At any design Xi' Thus the values of Aand (XI' xJ can be determined interactively Starting from the initial design = (2.0, 2.0) in 2 , the result obtained are shown in Table. Table Results Starting Values GO A -----, I Solution from Eq. ( 12.1 0 I ) I
Xl X 2 Xl
0.20000E + 01 0.20000 +01 -O.10oooE + 01 0.40022 E + 02 0.29289 E + 00 0.58579 E + 00 .: E .,. 00 0.58579E + 00 -O.10000E +01 0.31830 E + 02 0.1472E+ 00 0.5886E+OO 0.5886 E + 00 -0.10000 E + 01 0.26475 E + 02 0.86394 E - 01 0.69115 E + 00 J 0.10000 E + 00 0.69115 E + 00 --0. /0000 E +01 0.23898 E + 02 0.50714 E-OI 0.70102 E + 00 I 0.10000 E + 00 0.70102+00 -0.10000 E + 01 0.23846 E + 02 0.50011 E-Ol 0.70117 E +00 I 0.10000 E + 00 0.70117 + 00 -0.10000 + 01 0.23845 E + 02 0.50000 E - 01 0.70117 E + 00 ! 0.10000 E <-(10 0.70117 E+OO -O.1Ooo0E+01 0.23845 E + 02 0.50000 E - 01 0.701 17E + 00 I With lower bounds on X 1 and x 2 as 0.1 160 BC-2.4 012. Explain optimality criteria with a style displacement constraining. Ans: The optimality criteria methods are based on the derivation of an appropriate cliteria for specialized design conditions and developing a iterative procedure to find the optimum design. The optimality criteria methods were originally developed by Prager and his associates for distributed (continuous) systems and extended by Venkayya, Khot, and Berke for discrete systems. The methods were . first presented for linear elastic structures with stress and displacement constraints and later extended to problems with other types of constraints. We will presert the basic approach using only displacement constraints. Optimality Criteria with a Single Displacement Constraint Let the optimization problem be stated as follows: n Find X which minimize r(X)= L cixi (1) i= 1 Subjectto n (2) L --(=Ymax a. i = 1Xi Where C j are constants, Y.."" is the maximum permissible displacement, and a, depends on the force induced in member i due to the applied loads, length member i andYoung's modulus of memberi . The Lagrangian function can be defined as: L(X,A)= i cixi + ~ ( at. ymax) (3) i=l i=lx t At the optimum solution, we have dL ak ~ 1 dat :,;-=ck -A- 2 +A .. --;-=0, k=1,2, ...,n (4) oXk xk /=1 Xi OXj It can be shown that the last term in Eq. (4) is zero for, statically determinate as well as indeterminate structures so that Eq. (4) reduces to . ak (5) ck -A 2=0, k=1,2, ... ,n Xk or A=CkXr ak (6) Equation (6) indicates that the quantity C l X k 2 I a k is the same for all the design 161' BC-2.4