Date of Award
Master of Science
The objectives of this thesis are to explore the use of basic Statistical Process Control in a Wastewater Treatment Plant with secondary treatment. Additionally, to describe statistical behavior and review normality of some reportable parameters, also to analyze mainly parameters like BOD5 (Biological Oxygen Demand), Electrical Conductivity, and COD (Chemical Oxygen Demand) with seasons of the year and days of the week, using data from the Northwest (NW) and Robert R. Bustamante (RRB) Wastewater Treatment Plants from El Paso Texas.
The scope of this analysis is to understand how Statistical Process Control can help to predict peak and valleys of certain parameters, patterns of behavior, and find if the wastewater treatment show linear relationships to improve performance by anticipating adverse behaviors.
The methodologies used in this thesis are Non-experimental, longitudinal (trend), quantitative in nature. There are few statistical models used for Statistical Process Control and descriptive Statistics. That includes tools like Correlation Matrices, scatter plots, trend line insertion, review of the R2, time series graphs, and few tests using Capability Studies on a parameter to check capability index.
El Paso Water Utilities collected all the samples of wastewater and treated water, analyzed them using the methodologies, instrumentation, and other technical requirements as per the TPDES and EPA Agencies. EPWU collected water in regular intervals for twenty-four hours, creating a time-weighted average is part of their process. Both automated samplers and personnel from WWTP were involved in the sampling process. Performance of the Statistical Analysis involved SPSS® 19 (IBM SPSS® Statistics software) and Minitab 16, while Microsoft Excel became mainly a storage for database.
El Paso Water Utilities (EPWU) provided processed data used in this analysis. The period covered by the data was from January 1 2008 to December 31 2011 of parameters such as flows (incoming and outgoing) from the wastewater plants, including BOD5 and COD among others.
Outliers are included in the statistical analysis due to ethical reasons. There were no logbooks or records explaining the occurrence of special cases. This measure affected the Capability tests for most of the parameters since outliers were outside the upper limit. Based on the four rules applicable to control charts most of the process is statistically not controlled, thusly no Cpk assessment applies to them. A Cpk study using segments of timeline with varying results is part of the results.
The conclusions are that Statistical Process Control is applicable to the control of WWTP secondary treatment in the basic level. Furthermore, despite the process is continuous, mostly non-linear and dynamic, Statistics applied to Process Control described the process signaling the "out of control" alerts. All the variables tested are Statistically Normal fulfilling the basic requirements using Shapiro-Wilk as the Normality Test, except RRBInBOD5, and NWeffTHard.
The parameters reviewed are BOD5, COD, Electrical Conductivity, Total Suspended Solids, Total Phosphorus, TKN, NO2, and NO3. There are few points for the NO2 parameter. In addition, of all the explored linear correlations done in this study, there were few significant correlations found.
An analysis of Descriptive Statistics yielded interesting behavior patterns. Electrical Conductivity behavior patterns from RRB showed cyclic peaks during the fall of each year, EPWU WWTPs are non-linear systems, household wastewater and industrial wastewater are two different universes detected in the analysis, COD and BOD5 peak on several days of the week for each Wastewater Treatment Plant are instances discussed in this thesis.
The limitations are that lack of available time prevented from testing more Statistical Process Control tools to assess limitations of the Statistical Process Control. In addition to that, the temperatures used from NOAA were maximum and minimum both extreme temperatures. This is a limitation since the average does not represent the temperature that prevailed during the day. Applicability of results is for all parameters that met the normality requirement. Finally, analysis of the raw data collected by EPWU on a daily basis was not available at the time of writing up this thesis. The use of time-weighted averages on most of the parameters resulted in the loss of details that might have provided additional behavior patterns on the load arrival to the Wastewater Treatment Plant.
Received from ProQuest
Juan Carlos Adame
Adame, Juan Carlos, "BOD5 And COD Data Analysis From The Northwest And Robert R. Bustamante Wastewater Treatment" (2012). Open Access Theses & Dissertations. 1769.