About this lesson
Experiments and Design
Experiments are an inquiry process that is often used to support the design process. Experiments determine the relationship between independent factors or variables associated with the design or use of the item under investigation and dependent factors which indicate the performance of the item under investigation.
When to use
Experiments should be used when there is uncertainty about how the independent factors will impact the dependent factors. The data created by the experiments are used to model the relationship. If the relationship is already known, experiments are not needed. Experiments can assist in the design of new products/services/processes, upgrade of existing products/services/processes or when conducting problem solving analyses with products/services/processes.
The goal for experiments is to establish the relationship between the independent factors and the dependent factors. Therefore, the independent factors must be controllable and the dependent factors must be measurable. The experimental study is designed to create enough data to enable:
- Creation of a model between the independent and dependent factors,
- Expansion of knowledge about a product, service, or process,
- Establish a cause and effect between an independent factor and specific dependent factor condition or level,
- Optimization of the design of a product, service, or result.
Experiments are particularly helpful when changing a technology or design elements of a product, service, or process. In that case the experimental data provides valuable insight into likely performance of the changed system. Experiments are not needed when the design is not changing; either because it is being restored to a known configuration or the attribute under investigation has no performance impact.
Experiments have always been used by designers, but the experimental method has changed. The most common method is trial and error. The method taught in most technical schools as the scientific method is One-Factor-At-A-Time (OFAAT). Two statistically based methods are the full factorial Design of Experiments which determines the performance at the edges and corners of operation. The fractional factorial Design of Experiments statistically combines a much smaller number of experiments than OFAAT or full factorial DOE to determine the characteristics of the design space.
Hints & tips
- Most of the time, experiments will prove beneficial in understanding design performance.
- The selection of independent factors should be those that vary but are controllable by user/operators.
- The selection of dependent factors should be easily measurable and directly relate to the product, service, or process performance.
- Experiments can become very expensive, so plan your experiment approach to conduct the minimum needed to gain understanding about the issue under investigation.
Lesson notes are only available for subscribers.
PMI, PMP, CAPM and PMBOK are registered marks of the Project Management Institute, Inc.