You are on page 1of 5

MOW 227 Tutorial 1 (assessment)

7 August 2020
Student name Vaibhav

Student surname Parshutam

Student number 18207635

Use this fillable pdf form to capture all your answers for this assignment. Enter real numbers
using 4 significant digits, but note that all numbers (even integers) are displayed using 2 digits
after the decimal point. Save this file and submit it on ClickUP. These forms are graded
automatically. This assessment closes strictly on Tuesday 11 August at 9:00am. By
submitting your solution to this problem set on ClickUP, you confirm that you are aware of
the following Integrity Statement:
University of Pretoria Integrity Statement:
The University of Pretoria commits itself to produce academic work of integrity. I affirm that I
am aware of and have read the Rules and Policies of the University, more specifically the
Disciplinary Procedure and the Tests and Examinations Rules, which prohibit any unethical,
dishonest or improper conduct during tests, assignments, examinations and/or any other
forms of assessment. I am aware that no student or any other person may assist or attempt to
assist another student, or obtain help, or attempt to obtain help from another student or any
other person during tests, assessments, assignments, examinations and/or any other forms of
assessment.

Question 1 [5]

1. The stress at a critical location in a part is measured in service at each of your clients.
You have measured 1000 parts and the mean stress in the part is 250 MPa, with a
standard deviation of 40 MPa. If you measure the strength of the material that you use
to manufacture this part, the mean strength is 350 MPa, while the standard deviation
is 67 MPa.
What is the probability of failure? 0.100
[2]
How many of the 1000 parts are expected to fail? 100.00

2. You cannot accept so many failures, so you decide to improve the quality control during
manufacturing.
What should the standard deviation on the material
strength be such that you expect only 2% failures? 28.00 (MPa) [3]

Question 2 [25]
In this section, all the questions are from a company that makes available commercial opti-
mization algorithms (https://minlp.com/home). The main purpose is not to merely solve

Page 1 of 5
these problems. The solutions are even given on the website. The purpose of the home-
work is to solve the problems USING SOLVER IN EXCEL. So become familiar with the
Solver interface, so that you know how to set up a problem, how to add, modify or remove
constraints, and how to interpret the Lagrange multipliers once a solution is found.

1. Solve the following constrained optimization problem using Solver in Excel (from
https://minlp.com/mat-bar-example1). Use starting point [x1 , x2 ] = [2; 0]. x1 and
x2 are real numbers.

What is the optimal solution for the cost function? [1] -4.00
What is the optimal solution for x1 ? [1] 1.99
What is the optimal solution for x2 ? [1] 1.99
How many constraints are are active at the solution? [2] 1.00
What is the largest absolute magnitude of all the non-
-0.50
zero Lagrange multipliers? [2]
Predict the new cost function value if the active con-
straint with the largest absolute magnitude Lagrange -3.94
multiplier is relaxed by 0.1 unit. [2]
Re-solve the optimization problem by relaxing the dom-
inating active constraints by 0.1 unit. Report the new -3.94
cost function value. [2]

2. Solve the following constrained optimization problem using Solver in Excel (from
https://minlp.com/mat-bar-example3). Use starting point [x1 , x2 ] = [2; 0]. The
original problem requires x1 and x2 to be integers.

Page 2 of 5
Start of assuming x1 and x2 are real numbers. What is
8.15
the optimal solution for the cost function? [1]
What is the optimal solution for x1 ? [1] 5.31
What is the optimal solution for x2 ? [1] 2.72
How many constraints are active at the solution? [2] 2.00
What is the largest absolute magnitude of all the non-
4.34
zero Lagrange multipliers? [2]
Predict the new cost function value if the active con-
straint with the largest absolute magnitude Lagrange 7.70
multiplier is relaxed by 0.1 unit. [2]
Re-solve the optimization problem by relaxing the dom-
inating active constraints by 0.1 unit. Report the new 7.72
cost function value. [2]
Now add a constraint that x1 and x2 must be integers.
Now there are no Lagrange multipliers, and we cannot
use them to understand the sensitivity of the problem
w.r.t. the constraints. Resolve the problem. What is the
optimal solution for the cost function? [1]
What is the optimal solution for x1 ? [1]
What is the optimal solution for x2 ? [1]

Question 3 [20]
Optimize a 2 bar truss (similar to problem 2.2 in Arora, Chapter 2. But here we only design
the cross-sectional area instead of the inner and outer diameters of the tubes) shown in
Figure 1.

Figure 1: Two bar truss

Page 3 of 5
The 4 design variables are the height of the truss (h), the width of the truss (s), the area of
member 1 (A1 ) and the area of member 2 (A2 ). The external load W acts at an angle θ. The
angle θ can vary between 0° and 90°. The cost function is the mass of the truss structure:
p
m(h, s, A1 , A2 ) = ρ h2 + s2 /4(A1 + A2 ) (1)

where ρ is the density of the material. The constraints are that the normal stress in each
truss member must be less than some allowable limit σa :
F1 F2
σ1 = ≤ σa and σ2 = ≤ σa (2)
A1 A2
Finally the truss size is limited between reasonable lower and upper bounds: sl ≤ s ≤ su
and hl ≤ h ≤ hu .
Analysis: This problem is more realistic than the other problems in that all the required
values are not immediately available. Here we need to compute the forces in each truss
member F1 and F2 . Consider equilibrium of point 1, where the load is applied (see Figure 2).

Figure 2: Equilibrium of point 1

Equilibrium in x and y gives the two equations

ΣFx : −F1 sin(α) + F2 sin(α) − W cos(θ) = 0 (3)


ΣFy : −F1 cos(α) − F2 cos(α) − W sin(θ) = 0 (4)

These two simultaneous equations are written in linear algebra format as


    
− sin(α) sin(α)  F1   W cos(θ) 
  = (5)
− cos(α) − cos(α)  F2   W sin(θ) 

This allows us to compute the truss member forces F1 and F2 if we know the angle α (compute
this from h and s), the external load W and the angle at which it is applied θ.
Use constants W = 10000 N, ρ = 7.8 × 10−6 kg/mm3 , σa = 100 MPa, sl = 200 mm,
su = 1000 mm, hl = 200 mm and hu = 1000mm. Compute the solution of this optimization
problem (cost function and design variables). Since the truss has to be strong enough for all
possible angles 0° ≤ θ ≤ 90°, you need to consider many stress constraints simultaneously
(for each angle θ you regard as important). Consider the angles θ = [0; 30; 60; 90]° when
you compute the constraints.

Page 4 of 5
What is the optimal solution for the cost function in kg? [2] 0.36
What is the optimal solution for h in mm ? [2] 200.00
What is the optimal solution for s in mm ? [2] 332.59
What is the optimal solution for A1 in mm2 ? [2] 100.24
What is the optimal solution for A2 in mm2 ? [2] 78.21
At how many of the considered angles (0°, 30°, 60° and 90°)
does the stress in one of the members equal the maximum 2.00
allowable value ? [2]
If you select a stronger material (you make σa bigger), list the 3
largest Lagrange multipliers that inform you by how much you 0.00
can expect the cost function to improve. Be careful here: there
are probably many angles at which the stress in one of the 0.00
members equals the maximum allowable value (maybe some
members reach the maximum tensile stress and others 0.00
reach the maximum compressive stress) [3]
Predict the new cost function if you use a material for which σa
0.33 (kg)
= 110 MPa, using the Lagrange multipliers above. [3]
Re-solve the optimization problem for σa = 110 MPa. Report
0.33 (kg)
the new cost function value. [2]

Total: [50]

Page 5 of 5

You might also like