RACKGROlJND STUDY
2.3 Design of Experiment
bpenments are perfonned by investigators in virtually all fields of inquiry, usually to
discover something about a particular process or system. Liberally an experimcnt is a test. A
dcsigned experiment is a test or series of tcsL, in which purposeful changes are made to thc
input variables of a process or system so that we may observe and identify the reasons for
changes m the input responsc.
Gxperimental dcsign methods play an important role in process developmcnt and process
troublcshooting to Improve performance, The objective in many cases may be to develop a
robust proccss, that is, a process affected minimally by external sourccs of variability.
Applications of Experimental Dcsi!,'n:
Expenmental dcsign methods have found broad application in many disciplines. In fact,
expenmentalion as part of thc scientific method or proccss and as one of the ways wc learn
about how systems or processes work. Generally, we learn through a senes of activities in
,
which wc make conjectures about a proccss, perform experiments to generate data from the
process, and then use the infonnation from the experiment to estabhsh new conjectures,
which IcaU to new experiments, and so on.
Experimental design is a critically important tool in the engineering world for improving the
performance of a manufacturing proec,s, It a\loohas extensive application in the dcvelopment
of new processes The application of expenmental design techniquc~ early in process
development can result in-
• Improved process Y,ields.
• Reduced variability and eloser conformance to nominal or target requirements,
• Reduced development time
• Reduced ovcralll1me,
Experimental design methods also playa major role in engineering design acll~illes, where
new produetlo are developed and existing ones improved. Some application~ of experimental
design in engineering dcsign include-
• Evaluation and comparison of basic design configurations,
• Evaluation of material alternations.
• Selection of design parameters so that tne product will work well under a wide variety
of field conditions, that is, so that the product IS robust
• DeterminatIOn of Key product design parameters that impact product performance.
The use of experimental design in these areas can result in products that are easier to
manufatlun:, products that have enhanced field perlormance and reliability, lower product
cosl, ,md shorter product design and development llme.
Basic pnncip\cs
The three bu,ic principles of experimental design are replication, randomization and blocking, By replication we mean a repetition of the basic eJlpenment. RandomiTahon is the cornerstone underlying the usc of satisfied methods in eJlperimental deSIgn. By randomization we mean th,atboth the allocation of the experimental material and the order in which the individual runs or trials of the experiment arc to be perfonned arc randomly determined. Statistical methods require that the observations (or error:.) be independently distributed random variables. Randomization usually makes this assumption valid, By properly mndomiz:ing the experiment, we also a~slst in 'averaging out' the effects of e~traneous factors that may be present.
Blockmg ISa technique used to increase the precision of an experiment. A block is a portion of the experimental material that should be more homogeneous thun the entire set of material. Blocking involves making comparison among the conditions of interest in these experiment within each block.
These basic principles of experimental design are an important part of every experiment.
Historical Per,pechve.
The late sir Ronald A. fisher was the innovator in the use of statistical methods in experimental desib'll.For several years he was responsible for ,tatisties and data analYSISat the Rothamerted Agricultural Experiment station IIILondon, England. Fisher developed and first used the analYSISof variance as the primary method of statIstical analysis in expenmental design, In
1933,
fisher look a profe~sorship at the University of London. He lakr "as on the faculty of Cambridge Univers'ty and held visiting professorships at several un1vers,tiesthroughout the world. For an excellent biography of fisher, Su
.1.
F Box
(1978)
while fisher was clearly pioneer, there have heen many other significant contnbutors to the literature of experimental desib'll, meluding F. Yatee, R C. Bose, O. Kempthome, W. G.
Cochran, and G, E. P, Bon.
Many of the early applications of experimental design methods were In the agncultural and hlologieal science>and as a re~u1t,much ofthe termmology of the field ISderived from this heritage. However, the first industrial applications of experimental design began to appear in the 19303, initially in the British texllie and woolen industry. After world war-H, experimental deslj,'!]methods were introdoced to the chemical and process industries in the United Slates and Western Europe. These industry groups arc still very fertile areas ror using experimental dcsign in product and process development. The semIconductors and electronics industry has also used experimental design methods for many years with
considerable success.
III recent years, there has been veritable of interest in experimental design
10 the Umted Slates because many industries discovered that their off-shore compctl\ors have been usmg de"j,,"ed experiments for many years and that this has been an important factor m their competItive success. 1hc day is approachmg (hopefully rapidly) when all engineers WIll receive formal training in experimental desij,'llas part of their undergraduate education
GuidelinC!lfor Designing Experiments:
To ust: the statistical approach in designmg and analyzing an experiment, it is necessary thaI
everyone involved in the experiment have a clear idea in advanced of exactly what is to be
studied, how the data arc to be collected, and at least a qualitative understanding of how
these data arc to be analyzed. An outline orthe recommended procedure ISa follows:
