Western Constructions Case

Submitted by:
Savithri Anju
MS15A051
Case Summary:
The case falls into the category of a one dimensional cutting stock problem. The problem
statement is that the requirement for steel rods for current job order are of the below lengths
and quantity:

Length Requirement
45 382
55 410
65 296

Due to a limitation of the transporting truck the rods can be supplied only in lengths of 3m.
Hence the team needs to devise an optimum way of cutting the rods to meet the weekly
requirement. An additional point to be noted is that the requirement could vary from 30cm
to 80cm. Because of this the team needs to come up with an algorithm that simplifies finding
the patterns in which the cuts are to be made.
Mathematical Notations:
Let xi be the number of cuts made using pattern i. This is our decision variable for this
problem.
Formulation:
The objective that Mr.Pinaki Basu wants is to minimise wastage. As a first step we generate
the possible patterns such that using the wastage we cannot make another cut of the
required length. 18 such patterns exist as shown below:
6 0 0
5 1 0
5 0 1
4 1 1
4 2 0
3 3 0
3 0 2
2 3 0
2 0 3
2 2 1
2 1 2
1 4 0
1 1 3
1 2 2
1 3 1
1 1 3
0 3 2
[0 4 1]

777 cuts [0 3 2] The dual variable can be found using yB=1 . 18 ∑ ??1 = 382 ?=1 18 ∑ ??2 = 410 ?=1 18 ∑ ??3 = 296 ?=1 Results: Solving this as an LP gives the following solution: 92.44 cuts [4 1 1]. ?? ≠ 0 ??? ??? ? However another factor that needs to be considered is the fact that all the additional units that are made beyond the requirement is also a notional waste.The objective function can now be written as: 18 ??? ? = ∑(300 − (??1 ∗ 45 + ??2 ∗ 55 + ??3 ∗ 65)) ∗ ?? ?=1 Subject to the constraints: 18 ∑ ??1 ≥ 382 ?=1 18 ∑ ??2 ≥ 410 ?=1 18 ∑ ??3 ≥ 296 ?=1 And.074 cuts [3 3 0] and 101. 4. Incorporating this into the objective function would reduce the objective function to 18 ??? ? = ∑ ?? ?=1 And the constraints change from inequalities to equations.

. Similarly all patterns can be evaluated. On solving 5 feasible patterns were obtained and the total number of rods to be cut is reduced to 148.29. 4 of [3 3 0] and 101 of [0 3 2] Gives 380 of 45cm.667 with the first pattern.77 from the previous 198. Solving using column generation we assume the first three feasible patterns to be [6 0 0]. The basic matrix is 6 0 0 ?= 0 5 0 0 0 4 The solution to the dual yB = 1 is 1 1 1 ?=[ ] 6 5 4 The dual is formulated as a knapsack problem and solved using branch and bound technique. 4 3 0 ? = [?1 ?2 ?3] [1 3 3] = [1 1 1] 1 0 2 4 5 2 ?=[ ] 27 27 9 Using this dual solution the other patterns can be evaluated to understand if it is feasible or not. This still holds the upper bound of 199. 82 with the second pattern and 74 with the third pattern. For example consider the pattern [0 4 1]. For this scenario too the patterns are found and 32 patterns were identified. 4 5 2 0 26 [ ] [4] = 27 27 9 1 27 This is less than 1 so infeasible. The lower integer values of the cuts at the optimum 92 of [4 1 1]. The LP gives an upper bound to the IP and thus we can say that 199 is an upper bound to the IP. Using these patterns we get that we need to cut 63. [0 5 0] and [0 0 4]. The next part of the case states that Sumant says that having 4m rods instead of 3m can reduce the number of rods that need to be cut. 407 of 55cm and 294 of 65cm so we can have one sheet cut in [1 3 1] and one in [0 0 1].