By Mike Walmsley

       Most decisions we make in life are not complicated. We assess the pertinent information available to us, blend in a little common sense and intuition, and arrive at a direction to pursue. We are either successful in the outcome or not. If not, we learn.  On the other hand, decisions in the business sector are fairly complicated. Penalties for making an incorrect decision can cost us customer dissatisfaction, millions of dollars in lost capital, market share, and in the extreme case, our very business.

       The academic field of Operations Research has recognized this concern and has developed a separate branch of research known as the “Decision Sciences”. The original emphasis was placed on co-jointly developing techniques with the DoD (Department of Defense), DoE (Department of Energy) to address and identify risk management policy for potential safety hazardous situations in our environment and community (such as impacts due to nuclear, biological and chemical events). Later, the concepts were recognized to be applicable in the business sector.  While many of the techniques are fairly complicated in their mathematical reasoning, one such technique is not. It has been in existence and use since the late 1980s with a formal introduction cited in the text “Creating Innovative Products Using Total Design” * by Stuart Pugh (1996 Addison Wesley). Known as the “Pugh Matrix” technique, it is a favorite method for management teams dealing with problem solving in Multi-Concept, Multiple- Attribute situations.

     The output results of the process are product / process design concepts that are chosen and optimized based on open and documented criteria. Such a process forms well-understood product / process design values and forces cross-functional negotiation and buy-in. These criteria and their importance have been put in place by organizational consensus building and relate back to the driving customer wants and needs (CTQs). This is key to the Six Sigma DMAIC, DFSS process/product design improvement processes respectively. Because of this, the Pugh decision analysis matrix has become a key tool in the Six Sigma toolbox.

     The optimum use of the tool is for those situations where we have many alternative concepts to choose from, with each concept entailing multiple attributes (product / process considerations).

The process flow for tool use is:

1. From Six Sigma, Define the problem.
2. Define the cross-functional team that will work on the problem.
3. Define what we consider to be the “baseline”. e.g. a product or process that we have in place that is considered non-optimal.
4. Define the requirements (CTQs ~“attributes”) and criteria that will be used in the analysis.
5. Define the alternative concepts that will be included in the analysis.
6. Define rating scale that will be used in the prioritization and trade-off attribute analysis.
7. Define the scaling criteria that will be used in the concept analysis.
8. Set up the Pugh Matrix.
9. Analyze and rank each concepts benefit.
10. Conduct “what-if” simulation studies and re-assess concepts benefit (optional).
11. Select the concept(s) that detail the most benefit.
12. Initiate development/optimization analysis activities on the concept chosen.

As an example of how the tool can be used consider the following from the automotive industry. Antilock brake system Wheel Speed Sensors (3 / 4 per vehicle). 

Figure 1 The application and basic mode of operation

Mode of operation: Sensors are located at wheel ends (Hub or Hub and integral bearing). Sensor head is at most 2mm from ferromagnetic ring, integral to brake rotor. As wheel is turned, a field change is generated and a pulse is issued thru sensor head to vehicle ECU. Pulse width and frequency of pulses is compared to a tabled algorithm. This algorithm defines to a comparator in the ECU the speed at which a particular wheel is moving. All (4) wheels are monitored simultaneously. If a significant deceleration departure on one wheel is noted (indicating slip, lock), the vehicle anti-lock system is activated by the ECU.

The problem definition: When ABS systems first came out, the existing technology was deemed non-optimal to environmental temperatures, vehicle packaging, signal accuracy, resolution, high and low speed design limits, output signal level and waveform, Reliability, serviceability, packaging size (mass), EMI/EMC robustness, manufacturability, cost, and others.     The “Attributes”

Scope of analysis: Research at the time had indicated that there exist at most (14) separate design concept configurations that could be potentially pursued. The “Concept Alternatives”. Note that only one was in actual physical existence at the time (VR). Since the physical theory exists for the others, assessing the other concepts will require a great deal of design simulation, materials analysis and potential pricing with respect to the proposed environment and duty cycle.

Goal Statement: Find and select at most (3) sensor technologies that would entail the most benefit in terms of future development.

Solution: A team was formed, the technology concepts were researched and the initial Pugh Matrix was set up.

Figure 2 The initial Matrix and scoring criteria

Attribute rating criteria                                                             Concept Alternative scaling criteria

                                                 The concept Alternatives

                         Ranked Attribute           Scaled (zeroed)                          
                               importance                    baseline
 

How it works: To assess each of the concept alternatives, the concept alternative scaling criteria “sign” will be placed in front of the attribute importance for each attribute of a given concept alternative. E.g. If it were determined from research measurement that for Option #1 the attribute“ Piece cost comparison to existing system” were to be a benefit over the existing baseline system, then a “+5”  would be placed in that cell. The “5” coming from the importance of need for the attribute. If the particular research measurement were to show that a disadvantage would exist with respect to the existing baseline system, then a “- 5” would be placed in the cell. Each and every cell in the matrix would be assessed similarly.

Once all the cells are filled in, we would “sum” each column and look for those concept alternatives that were largest  from a positive standpoint over all the others. A “larger the better” scenario.

Each member of the team was assigned research tasks that required obtaining meaningful variables measurements for each of the key Attributes.

Obtaining the measurements required a period of two months to accomplish. Once the measurements were obtained, they met in a team meeting and completed the matrix.

          Figure 3 The completed matrix (and solutions identified).