Charting and Evaluation of Real-Time Continuous Monitoring Water Bioburden

Area of Science:

• Environmental Science
• Statistical Process Control
• Microbiology

Background:

• Continuous monitoring of municipal water bioburden generates high-frequency data.
• Laser-induced fluorescence (LIF) analyzers produce positively autocorrelated data streams.
• Autocorrelation in data complicates interpretation and increases false alarms on Shewhart control charts.

Purpose of the Study:

• To analyze municipal water bioburden data to minimize autocorrelation.
• To improve the charting and evaluation of monitoring process behavior.
• To establish when to take action to maintain the process under statistical control.

Main Methods:

• Application of statistical analysis to high-frequency bioburden data.
• Utilized model-based and model-free methods to address data autocorrelation.
• Compared skip sampling, time series models, and batch means control charts.

Main Results:

• Skip sampling and time series models successfully removed autocorrelation.
• Batch means control charts, a model-free method, were favored for practical application.
• Averaging successive data points into batches reduced data frequency and improved interpretability.

Conclusions:

• Batch means control charts, with practically determined limits, convert high-frequency data to low-frequency data.
• This method results in standard control charts with fewer false alarms, revealing underlying process trends more clearly.
• The approach is applicable to environmental monitoring of inert particles and microbes in controlled environments.