AXIOMATIC DESIGN IN SOFTWARE DESIGN FOR SIX SIGMA (DFSS) - Software Design for Six-Sigma: A Roadmap for Excellence - page 333

over the design. It also implies a high level of design quality and reliability because
the interaction effects between the FRs are minimized. In addition, a failure in one
(FR, DP) combination of the uncoupled design matrix is not reflected in the other
mappings within the same design hierarchical level of interest.
For the decoupled design, the path independence property is somehow fractured.
As depicted in Figure 13.2(b), decoupled design matrices have a design settings
sequence that needs to be followed for the functional requirements to maintain their
independence. This sequence is revealed by the matrix as follows: First, we need to
set FR2 using DP2 and fix DP2, and second set FR1 by leveraging DP1. Starting
from setting (1), we need to set FR2 at setting (2) by changing DP2 and then change
DP1 to the desired level of FR1.
The previous discussion is a testimony to the fact that uncoupled and decoupled
designs have a conceptual robustness, that is, coupling can be resolved with the proper
selection of DPs, path sequence application, and employment of design theorems
(El-Haik, 2005).
The coupled design matrix in Figure 13.2(c) indicates the loss of the path indepen-
dence resulting from the off-diagonal design matrix entries (on both sides), and the
design team has no easy way to improve the controllability, reliability, and quality of
their design. The design team is left with compromise practices (e.g., optimization)
among the FRs as the only option because a component of the individual DPs can be
projected on all orthogonal directions of the FRs. The uncoupling or decoupling step
of a coupled design is a conceptual activity that follows the design mapping and will
be explored later on.
An example of design coupling is presented in Figure 13.3 in which two pos-
sible arrangements of the generic water faucet 2 (Swenson & Nordlund, 1996) are
displayed. There are two functional requirements: water flow and water temperature.
The Figure 13.3(a) faucet has two design parameters: the water valves (knobs) (i.e.,
one for each water line). When the hot water valve is turned, both flow and temperature
are affected. The same would happen if the cold water valve is turned. That is, the
functional requirements are not independent, and a coupled design matrix below
the schematic reflects such a fact. From the consumer perspective, optimization of
the temperature will require reoptimization of the flow rate until a satisfactory com-
promise amongst the FRs, as a function of the DPs settings, is obtained over several
iterations.
Figure 13.3(b) exhibits an alternative design with a one-handle system delivering
the FRs , however, with a new set of design parameters. In this design, flow is adjusted
by lifting the handle while moving the handle sideways to adjust the temperature. In
this alternative, adjusting the flow does not affect temperature and vice versa. This
design is better because the functional requirements maintain their independence per
axiom 1. The uncoupled design will give the customer path independence to set either
requirement without affecting the other. Note also that in the uncoupled design case,
design changes to improve an FR can be done independently as well, a valuable
design attribute.
2 See El-Haik, 2005: Section 3.4 for more details.