OSHA requires that employees who are subjected to an 8-hour time weighted average (TWA) noise exposure of more than 85 dB must be enrolled in a hearing conservation program. Because of insufficient monitoring and potentially a lack of solid data analysis, many employers place their entire work force in such a program. The practice is an admission that all employees are being exposed to hazardous noise levels, when in fact many are not. It also is a blanket administrative control and reduces the focus of effort on improving and eliminating the sources of high noise levels.
Instead of this blanket approach, the Six Sigma DMAIC roadmap can be employed to properly measure the noise exposure of the employees and prioritize the best approaches for reducing noise levels. The goal is to maintain levels below the OSHA limits and reduce the need to enroll employees in the hearing conservation program.
Following the Roadmap
W.R. Grace & Co. is a specialty chemical manufacturing company based in Columbia, Md. It has manufacturing sites across the globe and employs more than 6,000 people. Grace's manufacturing site at Curtis Bay (Baltimore) operates four separate plants and employs 400 hourly and 200 salaried workers. All employees at the site are enrolled in the hearing conservation program. Audiometric tests are completed annually by the medical department and a history is maintained for each employee.
In an effort to improve working conditions at the site, a team was established to analyze the conditions in one of the manufacturing plants. The team was composed of plant personnel, a representative from the Curtis Bay safety department and a Six Sigma Black Belt. An industrial hygienist from Grace's corporate location also was used as a resource.
The team followed the Six Sigma DMAIC roadmap. DMAIC stands for Define, Measure, Analyze, Improve and Control. All Six Sigma efforts start with a practical problem. The next step is to translate this into a statistical problem by finding baseline data or creating a quantifiable, measuring system for the key inputs and outputs. From the statistical analysis, priorities are set and a statistical solution is developed. Finally, the numbers are translated into a practical solution and implemented through the use of a solid control plan.
With this methodology as the guide, the team reviewed the existing operations in the plant. They consisted of three separate production lines and one warehouse operation. A total of 11 operators per shift ran the plant during each of three daily shifts. Process mapping was used to list all of the key process output and key process input variables (KPOVs and KPIVs). The outputs were concerns such as the noise exposure levels, safety, production rates, profitability and cost of hearing loss to employees and the company. The KPIVs consisted of the 11 individual job responsibilities, the shift and the equipment in the plant.
Cause and effect matrices took the information from the process map and prioritized the KPIVs in order of their impact on the noise levels for the operators. This was done for each job in the plant. In addition to the C&E matrix, historical noise exposure data was reviewed through the use of I-MR control charts. The statistical and visual information provided by these charts allowed the team to see how certain jobs varied and provide a new way to look at the problem of noise exposure.
For example, Chart 1 is for an operator who averages 78 dB on an 8-hour shift. Note that the variation in noise exposure was high, indicating that the individual performing this job is constantly moving in and out of areas with different noise levels. Chart 2, however, indicates a very different situation. For an 8-hour shift, this operator was exposed to an average of nearly 86 dB, with a much lower standard deviation than the first one. From the chart, it appears that around points 125-150 the operator was in a much quieter area, more than likely a meal break.
The qualifying information provided by the process maps, cause and effects matrices and failure modes and effects analysis when used in accordance with the statistical analysis (I-MR charts, basic statistics and analysis of variance) lead to a conclusion that the 11 jobs in the plant should not be grouped together in a hearing conservation program because each individual has very different exposures to noise. By using analysis of variance, it was easy to see a statistical difference among the jobs. From this information the team divided the operators into similar exposure groups or SEGs.
Through the define, measure and analysis steps of the roadmaps, the team was able to list the key sources of variation in noise exposure. From this, the team developed a sampling methodology that would provide the statistical information necessary to conclude, at a 95 percent confidence level, that an individual was or was not being exposed to an 8-hour noise TWA of 85 dB or more. A plan was developed using nested analysis of variance. The key sources of variation to be analyzed were the SEG, in this case each of the jobs, the shift (7-3, 3-11 or 11-7) and the dosimeter being used to measure the levels. A sampling plan was created using 18 or 24 samples to account for the variation and provide the necessary statistical significance. The more samples taken the higher the level of confidence, but the more difficult to complete in the plants. Chart 3 below shows the structure of the sampling plan.
Once the data had been collected, it was analyzed using Minitab or a similar statistical software package. It was determined that 75 percent of the variability was attributable to operator responsibilities. This helps to verify the division of SEGs. The shift was not a significant variable in the model. The dosimeter represented just over 23 percent of the process variation. Based on the Six Sigma rule of thumb, if the measurement system represents less than or equal to 30 percent, it is adequate for the process and not a critical area for process improvement.
Based on the measurements taken, it was possible to conclude that only three of the 11 operators were potentially exposed to an 8-hour TWA of greater than 85 dB. The results of the nested analysis of variance provide an average and an upper and lower control limit at the 95 percent confidence level. If the SEG (operator job) had an upper control limit of greater than or equal to 85 dB, the group would be enrolled in the hearing conservation program. If not, the group would not have to be included in the program.
The method provides statistical support for management decisions. The information provided during the completion of the DMAIC roadmap also provides a means to prioritize improvement projects. The failure modes and effects analysis provides a living document and action plan to make changes. In the past, the plant supervisors would consider posting Hearing Protection Required signs or purchasing noise abatement equipment. This roadmap adds other options such as hard-wiring interlocks to turn off loud equipment when not required, modifying job responsibilities or ergonomic alterations to provide relief from noise. The sampling plans also provide a means to quantify the effects of the improvements and to reprioritize.
The use of the Six Sigma method is applicable to other areas of environmental, health and safety. The roadmap provides the data required to make the best decisions. The control phase requires work to be assigned and completed. The tools used also provide the documentation and means to follow the success of the improvements.
Donald L. Cain, P.E., is a Six Sigma Master Black Belt with W.R. Grace & Co. He has a bachelor's degree in mechanical engineering from Villanova University and a masters's degree in business administration from Loyola College of Maryland.