EXERCISE 6: CURVE FITTING (REGRESSION)
1. An engineer works in the Research and Development (R&D) unit for one of the
largest steel manufacturers in Malaysia. He is tasked to conduct an experiment to investigate
the percent elongation of a newly developed steel truss design for roof support when heat is
applied on it. Table 1 shows the data collected from the experiment.
Table 1: Percent Elongation of Steel Truss at Various Temperatures
Temperature, T (˚C) 200 250 300 350
Percent Elongation, E
7.0 8.2 9.1 10.0
(%)
Upon completion of the experiment, the engineer is required to estimate the percent
elongation of the steel truss at 280 ˚C.
(a) Since the data in Table 1 shows the percent elongation increases with temperature, use
the Linear Regression method to obtain an equation of percent elongation, E as a function
of temperature, T in the form of E(T) = A + BT where A and B are constants. Then, help
the engineer to estimate the percent elongation of the steel truss at 280 ˚C.
(b) Calculate the correlation coefficient, r and standard error of the estimate, Sy/x for the
equation obtained in (a).
2. Line stability index (LQP) is used to determine the voltage stability of a power system
network. The index value of 1.0 indicates that voltage collapse starts to occur in the system.
In a study to observe the impact of solar output power, P (in p.u.) to the voltage stability of a
network, the LQP data is plotted as shown in Figure 1.
1.4000
1.2000 1.1130
1.0000
LQP index 0.8000
0.6000
0.3500
0.4000
0.2000 0.0513 0.0103
0.0000
0.36 0.72 1.08 1.44 1.8 2.16
-0.2000
Solar output power (p.u.)
Figure 1: Impact of solar output power to voltage stability
(a) Use the quadratic regression method to represent the data of LQP index over the solar
output power in a quadratic equation.
(b) From the equation found in (a), use 3 iterations of Bisection method to predict the amount
of solar output power that causes voltage stability of the system starting to collapse.
Given that the initial value of Pl = 1.0 p.u. and Pu = 1.8 p.u. Find the approximate percent
relative error, |εa| for each iteration.
