Utilizing Quantum XL and Plackett-Burman Design for Efficient Multifactorial Product Design in Servo-Hydraulic Systems
Explore how Quantum XL and Plackett-Burman DOE optimize servo-hydraulic power unit design by screening 11 variables, improving model accuracy, and minimizing power loss with advanced regression and response surface methods.
Utilizing Quantum XL and Plackett-Burman Design for Efficient Multifactorial Product Design in Servo-Hydraulic Systems
1.
Utilizing Quantum
XL forProduct
Design with
Multifactorial
Parameters
Plackett-Burman Design for Development
of Servo-hydraulic Power Units
by Ramon Balisnomo
August 2, 2026
2.
History Behind ScreeningDOE
• The British Ministry of Supply asked Robin Plackett and
Peter Burman to develop a method for efficient learning
under severe experimental constraints. Their goal was to
study many variables with very few experimental runs.
• They assumed interaction effects were negligible,
allowing focus on main effects only.
• Oxford University Press published their journal The Design
of Optimal Multifactorial Experiments in June 1946.
3.
Design
Objective
Identify the designvariables
that most significantly impact
system efficiency or pressure
ripple during the early
development of a servo-
hydraulic power unit. The
coefficient of determination
must be at least 85 percent.
Thought Process Mapfor Conducting Experiments
star
t
stop
K > 5
Use Screening Design
(i.e., Plackett-Burman)
R2
≥
85%
Use Factorial or
Response Surface
Design
No
No
Yes
Yes
K=number of factors
23.
Rationale Behind 2nd
DOEfor Servo-hydraulic Power
Loss
The design team aims to better
understand power loss. The three factors
previously identified from the screening
DOE are:
• System Pressure
• Case drain / return back pressure
• Internal clearance (leakage
clearance)
Summary
• The Plackett-Burman(PB) design required only 12 runs to
evaluate 11 factors, while a full-factorial design would have
needed 2,048 runs (2number-of-factors
).
• In three out of the four cases studied, the main effects were
sufficient for accurately predicting the critical quality
performance metric.
• The PB design eliminated 8 out of the 11 factors that were
suspected to influence Power Loss.
• The coefficient of determination for the Power Loss model
increased from 66% to 99% when a response surface design
was used.