Sunteți pe pagina 1din 4

DEPARTMENT OF CHEMICAL TECHNOLOGY

HCT CHEMICAL TECHNOLOGY

NAME : MASCOT KAPETA

REG NO. : R188880P

LEVEL : 2.1

LECTURER : MRS ZINYAMA

MODULE CODE : CT213

MODULE TITLE : RESEARCH METHODS

AND STATISTICS

ASSIGNMENT : 1
Scientists and engineers often perform experiments which either vary one factor at a

time or varying multiple factors at once. According to (Wahid, 2013:80) “ Varying

one factor at a time is a method of designing experiments involving the testing of

factors, or causes, one at a time instead of multiple factors at once.” (Sheen, 2002:102)

argues that varying multiple factors at once, “…varies several factors simultaneously

and more efficient when studying two or more factors.”

The one factor at a time approach can be illustrated with an example by (Hunt, 2008)

in which he illustrated that supposedly we have a process with two inputs and one

output (Y). What could be done is to vary the inputs one at a time and see their effects

on the output. All inputs are held but one fixed, and the best result noted. Fix the

second input at that best value, and so on until we run out of input factors. Another

example is in the case of a car breakdown as propounded by (Owel, 2005). He says

when fixing the car, the mechanic fixes any possible part he or she thinks is the

problem and starts the car. If the car doesn’t start, the mechanic attends to another part

of the engine and starts the car. If the problem persists he repeats the process until the

car starts. So in this example, every stage is a one factor at a time approach. The car is

fixed stage by stage until the problem is fixed. The advantages of this approach are

that it is favored by non-experts, especially in situations where the data is cheap and

abundant and it might be useful if the real problem has too few experimental units to

do anything more. However, the disadvantages of this approach are that the method is

grossly inefficient leading to unnecessarily large number of experimental runs. In the

example of the mechanic, the mechanic should have connected the engine to a

computer and run the diagnosis instead of doing unnecessary runs. Another

disadvantage is that the experimenter is unable to study interactions among the factors.

(Owel, 2005:80) further explains “Interaction between causes is not measurable due
to no information which can mislead the optimal circumstances of the process.

Furthermore, each trial is run by comparing the detected outcome with prior results

which make the estimation of the factor effects less precise.”

Varying multiple factors at once is a more effective way to determine the impact of

two or more factors on a response. (Wahid, 2013) postulates that varying multiple

factors at once is a powerful tool and beneficial where traditional engineering analysis,

verification, and simulations are challenging to obtain. A supportive example of this

approach is illustrated by (Frey, 1998) of an engineer in planning an experiment in a

rubber glove production in Malaysia. The objective was to determine the effects of 8

variables on several important dimensions of rubber gloves. Fractional factorial was

employed. The experiment was replicated twice, resulting in a total of 32 runs. All of

the runs were conducted in a random order. Significant interaction was found

between the main controllable factors, latex temperature factor (B) and oven

temperature after coagulant dip factor (G). This interaction is important. The rest of

the factors’ interaction were found. Without this approach it would have been difficult

to find interaction between these factors. The advantages of this approach are that it

requires far fewer tests than one factor approach for valid results. This means less

resources for the amount of information obtained and crucial factor for industry. The

interactions among the factors are detected, which are considered for the experiment

and one factor at a time approach cannot find them,(Sheen, 2002) . (Hunt, 2008) adds

that the approach is simpler, more efficient and will need fewer experimental runs to

examine the impact of two or more factors on a response of interest. However the

disadvantages are that it might be costly as it requires extra resources for factor

manipulation at once.
References

Frey, G. (1998) Improvement of one factor at a time through design of experiments,

World Applied Sciences. Brazevia: World Applied Sciences. doi:

10.5829/idosi.wasj.2013.21.mae.99919.

Hunt, J. (2008) The Experimenter’s Companion. 2nd ed, The Writing Studio. 2nd ed.

Cape Town: ASQC Quality Press.

Owel, C. (2005) “RSM Simplified: Optimizing Processes Using Response Surface

Methods for Design of Experiments,” American Statistician, 53(2), p. 80. doi:

10.1080/00031305.1999.10474445.

Sheen, A. (2002) Six Sigma for Electronics Design and Manufacturing. McGrall-Hill:

Weirlder.

Wahid, P. (2013) Training for Design of Experimements. First. Crompton: Work

Study.

S-ar putea să vă placă și