Applied Statistics for Engineering

To remain competitive, companies continually strive to enhance product quality through efficient designs and improved manufacturing processes. Systematic analytical studies by engineers and physical scientists are essential in process improvement efforts. Statistical tools - including carefully planned experimental designs, proper sampling techniques, innovative data analysis, and correct data interpretation - are key elements in developing such improvements while saving time and money.

Southwest Research Institute (SwRI), one of the world's leading research and development organizations, has successfully applied statistical approaches to technical problems for its many clients around the world. Services include:

• Experimental Design
• Statistical Methodology Applications and Data Analysis
• Statistical Process Control
• Regression and Response Surface Modeling
• Reliability and Failure Analysis
• Sampling Plan Development
• Statistical Consulting and Training

Experimental Design

The Institute designs experiments that require a minimal number of experimental runs while maximizing the usefulness of the data. Instead of testing all combinations of design conditions, selected subsets are used to efficiently quantify the relationships between the test factors. These smaller designs employ strategies based on statistical methods as well as proven science and engineering techniques, minimizing the time and cost of the experiments.

Using experimental design techniques, the Institute evaluates the exhaust emissions and costs associated with several operating conditions in natural gas locomotives.

Statistical Methodology Applications and Data Analysis

Using a variety of analytical techniques, SwRI statisticians assess the relative effects of the various factors studied in an experiment. Aided by statistical software packages that process data, these procedures help identify and quantify the variation sources in an experiment.

Statistical Process Control

To help manufacturers identify and isolate major sources of product variation, the Institute develops statistical methods to monitor and control processes. Statisticians have applied these quality improvement tools in a variety of applications to develop several innovative approaches for statistical process control.

Regression and Response Surface Modeling

Predictive models, such as those based on regression analysis, are important results of experimental programs. Using graphical techniques, models help experimenters visualize complex interrelationships. Models can also identify and isolate the experimental conditions associated with the maximum or minimum response of a system.

Response surface modeling helps determine how input torque and speed influence torque loss in various transmissions.

Reliability and Failure Analysis

Reliability assessment studies are performed in a variety of applications using a combination of statistical and mechanical evaluations. These techniques determine the frequency of fatigue cracks, the mean time to failure of individual components, and the reliability of manufactured products.

The Institute evaluated fatigue cracking in space shuttle auxiliary power unit turbine wheels. To reliably estimate the useful life of a wheel, a team of statisticians and engineers developed a working hypothesis for potential wheel cracking.

Sampling Plan Development

Developing a useful sampling plan to collect necessary data is vital to experimental programs. Institute statisticians are specialists in this complex process, particularly in analyzing and interpreting survey results. A variety of sophisticated statistical procedures and software are utilized to obtain the most information from sample data.

Statistical Consulting and Training

SwRI regularly consults with a variety of clients on problems requiring applied statistical analysis. These problems often involve quality improvement issues, and require knowledge and experience in these activities. Topics typically include:

• Industrial experimental design
• Regression analysis and response surface modeling
• Statistical process control
• Reliability and risk assessment
• Graphical techniques and applications
• Total quality management

In addition to having received several departmental audit approvals, all departments in the Fuels and Lubricants Research Division have achieved certification to ISO 9002, the "Model for Quality Assurance in Production and Installation," and accreditation to ISO/IEC Guide 25, "General Requirements for the Competence of Calibration and Testing Laboratories."

Fuels and Lubricants Research Brochures
SwRI Brochures SwRI Publications SwRI Home