Вы находитесь на странице: 1из 7

Residuals and Biosolids 2008

USE OF QUALITY SIX SIGMA TOOLS TO OPTIMIZE SLUDGE DEWATERING BY CENTRIFUGATION Mikel E. Goldblatt, P.E., Gold Star Water Consultants, Inc. 301 Holmecrest Road Jenkintown, PA 19046

ABSTRACT The solid bowl centrifuge has emerged as a dewatering machine of choice in many treatment plants processing biological or mixed sludges. The high G forces promote the formation of highly compacted, dry sludge cake. Other advantages include a small footprint, and an enclosed system that reduces odor. The operating costs include maintenance of a precision-balanced high speed machine, cost of electrical power, and flocculant aids (polymers). This paper will demonstrate how Quality tools from the Six Sigma body of knowledge can be employed to optimize overall sludge processing costs when using a centrifuge. Perhaps many operating centrifuges could increase reliability and reduce overall costs by 25% or more, using these techniques! KEYWORDS Dewatering, sludge processing, centrifuge, optimize, Quality tools, Six Sigma INTRODUCTION Two Quality tools from the Six Sigma and many previous bodies of knowledge are highlighted. The first tool presented is the Quality Function Deployment (QFD) tool, which is a cause-andeffect and weighting prioritization tool. Next, the Design of Experiments (DoE) tools are illustrated, with sample data and a demonstrated optimization of a sludge dewatering centrifuge operation. Table 1 categorizes basic centrifuge parameters, assuming some parameters are set and unchangeable (for a particular unit), some are changeable but not easily controllable, and others are controllable variables. For clarity, the sludge being fed to the centrifuge will be referred to as slurry or sludge slurry and the dewatered sludge leaving the centrifuge as the cake or sludge cake.

Copyright 2008 Water Environment Federation. All Rights Reserved

750

Residuals and Biosolids 2008

Table 1: Categorization of Parameters for Solid Bowl Centrifuge Sludge Dewatering Set Parameters Unit size, brand, type RPM, G-force Uncontrolled parameters Slurry percent solids Slurry dewaterability primary:secondary ratio compactability Adjustable variables Slurry feed rate to centrifuge Polymer Flocculant Treatment Type, Dilution, Dosage Mixing energy and extent of mixing of polymer and sludge Scroll differential speed

Pond height (for this exercise well assume it is set)

QUALITY FUNCTION DEPLOYMENT TOOL Next, we want to further organize the parameters in terms of big Ys, little ys, and Xs. The big Y is the goal. In this case we want to minimize the big Y, Sludge Processing Cost. We can use the Quality Function Deployment (QFD) spreadsheet tool to organize and sequence what elements are needed, and how parameters can be manipulated to reach the goals. In short, the purpose of the QFD is to determine the Hows needed to yield the Whats. We list the big Y as the goal of minimizing overall sludge processing cost while maintaining reliable operations. Then we list the little ys including Cake disposal cost (dependent on cake dryness) Electrical power cost Labor and maintenance costs Reprocessing centrate solids cost (dependent on clarity of centrate) Flocculant polymer cost These data go into the QFD matrix, with weighting factors associated with each item and relationship factors between items. Figure 1 shows a filled out QFD matrix. The Ys, or dependent variables, appear in the wide column to the left, starting with Sludge Cake % Solids. Just to the left of each item is a numerical weighting of importance of that item. These values are reached by consensus of experienced operations personnel, and will certainly vary from case to case. In this case, a weighting factor of 9.0 has been assigned to maximizing Sludge Cake % Solids. Other items are given weighting scores, so that a relative percentage weight is assigned to each item in the next column to the left. The X categories appear across the top, starting with Feed Slurry Solids. Next, consensus relationship values are assigned. These appear in the boxes, relating each X with each Y where applicable. The ratings are as follows.
Copyright 2008 Water Environment Federation. All Rights Reserved

751

Residuals and Biosolids 2008

circle with center dash = strong relationship, score of 9 circle = moderate relationship, score of 3 triangle = weak relationship, score of 1

Once all these relationships are determined and recorded, a score for each X is tallied, based on the quantification of the judgments made. Finally, at the bottom of the chart a calculated relative importance of each X item appears. We can measure but we cannot control Feed Slurry Solids (at the centrifuge feedpoint), or Feed Slurry Dewaterability Ranking. We can control Feed Slurry Flow Rate and Scroll Differential Speed, as well as Polymer type, dilution, dosage and polymer/slurry mixing. These controllable parameters make up almost 2/3 of the relative importance, as calculated from the judgments above and illustrated in the QFD. Devising a series of plant tests to optimize these parameters leading to highest cake solids and highest solids capture should be feasible and have a high probability of yielding a quantifiable transfer function. Figure 1: Quality Function Deployment Tool

Copyright 2008 Water Environment Federation. All Rights Reserved

752

Residuals and Biosolids 2008

DESIGN OF EXPERIMENTS The Design of Experiments (DoE) tool is used to define sets of plant tests where parameters are varied in a planned and deliberate manner. The method allows variation of multiple parameters in a manner that maximizes useful information derived from a minimal number of tests. 1. The first step is to determine which results are targeted, then decide which variables can be manipulated to achieve the desired results. The QFD exercise yielded information indicating the importance of polymer choice, feedrate, dilution and mixing. Other parameters such as differential scroll speed came up with a low score in the QFD, though we know that is important. We could circle back and reconsider the QFD weights assigned. However, in this case let us assume that the machine is put on automatic control based on torque, and that the controller will adjust scroll speed optimally. Further, we assume that this variable will not cloud the results attained by manipulating the polymer parameters. 2. Next, we plan a wide-reaching set of experiments, wide enough to capture all suspected operable areas and failure points or limits. We can use a factorial design, a fractional factorial design or other method to maximize coverage of parameter manipulation while minimizing the number of tests required. In this case we choose polymer dosage, polymer/slurry mixing energy and polymer dilution as the three manipulable parameters. Perhaps we will have two polymers to choose between. This gives us four variables. If we limit to two choices per variable, and perform a full factorial set of experiments, we have to perform 16 tests to cover the 24 choices one time each. Of course we would want to perform repeat tests to minimize the effect of noise or unaccounted for variation on the results. Figure 2 shows a copy of a table taken from Minitab, Inc. software with green, yellow, red color-coding indicating judgments to go (green), proceed with caution (yellow) or reconsider (red). The codes aid the experimenter in determining the minimum number of tests to perform while minimizing the probability of confounding the possibility of mis-assigning causes to the wrong variable or set of variables. Figure 2: Available Factorial Designs (from Minitab, Inc. software)

Copyright 2008 Water Environment Federation. All Rights Reserved

753

Residuals and Biosolids 2008

3. Let us drop down to 3 factors and run a full factorial design. We will first do some preliminary screening, then choose a polymer dilution to hold. That leaves Polymer choice, polymer/slurry mixing energy, and polymer dosage. Eight tests, twice or thrice each means 16 or 24 tests will be run. Table 2 shows the factors and definitions and Table 3 shows the full factorial test cases. Table 2: Factors and Definitions
Factor A B AB C AC BC ABC Factor Name Polymer Dosage Polymer X Dosage Mixing Energy Polymer X Mixing Energy Dosage X Mixing Energy Polymer X Dosage X Mixing Energy Level 1 L Low Low Low(-) Level 2 H High High High(+)

Table 3: Full Factorial Design of Experiments, Three Variables, Two Values Each
Trial 1 2 3 4 5 6 7 8 Polymer + + + + Dosage + + + + Mixing + + + +

4. We run the tests, taking the data on sludge cake solids and centrate solids at each condition, then use statistical software to plot the main effects and interactive results. Table 4 shows results for one set of tests. Table 4: Full Factorial Test Results, Percent Solids in Sludge Cake
Trial 1 2 3 4 5 6 7 8 Polymer + + + + Dosage + + + + Mixing + + + + Sludge Solids 31% 27% 32% 29% 30% 30% 31.5% 31%

5. The results, sludge cake solids as a function of polymer choice, polymer dosage and mixing energy, and combinations of the factors are then plotted. Figure 3 shows the three Main Effects plots, where each graph represents the average response for a single factor.
Copyright 2008 Water Environment Federation. All Rights Reserved

754

Residuals and Biosolids 2008

On the leftmost chart one can observe that the polymer designated Low yielded a higher sludge cake solids level, on average, 31.13%. On average, the High dosage yielded higher solids cake (30.9%), and on average the High mixing energy yielded a slightly higher solids cake (30.6%). Figure 3: Main Effects Plots
Polymer
32.00 31.00 30.00 29.00 28.00 27.00 Low High 29.25 31.13 32.0 31.0 30.0 29.5 29.0 28.0 27.0 Low High 29.0 28.0 27.0 Low High 30.9

Dosage
32.0 31.0 30.0

Mixing Energy

30.6 29.8

6. Figure 4 shows two-way interactive plots. Where plots are parallel, there is no interaction between the factors. Where there is some perpendicularity, interaction is indicated. Observation of the combined plots shows that there is a significant interactivity of the two factors, polymer and mixing energy. One polymer gave unfavorable results with high mixing energy whereas the other polymer seemed barely effected by the difference in these two mixing energy points. Table 5 shows the relative effects of the factors (variables) and the combinations of variables. Numerically, Polymer X Mixing Energy is shown in the table to have a relatively large effect on the outcome. Figure 4: Interaction Plots
Polymer X Dosage

Dosage Low Dosage High

Polymer X Mixing Energy 32.0 31.5 31.0 30.5 30.0 29.5 29.0 28.5 28.0 27.5 27.0

32.00 31.50 31.00 30.50 30.00 29.50 29.00 28.50 28.00 27.50 27.00

31.8 30.50 30.0

31.5 30.8 30.5

Mixing Energy Low Mixing Energy High

Dosage X Mixing Energy 32.0 31.0 30.0 29.0 28.0 30.0 29.0 31.3 30.5

Mixing Energy Low Mixing Energy High

28.50

28.0 27.0 Low Low Polymer High Dosage High

Low Polymer

High

Copyright 2008 Water Environment Federation. All Rights Reserved

755

Residuals and Biosolids 2008

Table 5: Degrees of Freedom, Sum of Squares and Relative Effects


Anova Source Factor Polymer Dosage Mixing Energy Polymer X Dosage Polymer X Mixing Energy Dosage X Mixing Energy Polymer X Dosage X Mixing Energy df 1 1 1 1 1 1 1 SS 7.0 3.8 1.5 0.0 5.3 0.0 0.3 Effect -1.875 1.375 0.875 0.125 1.625 -0.125 -0.375

7. Based on the results in the first set of tests, another DoE can be set up where the ranges of the variables are reduced, or midpoints or several points along the high/low curve for one variable can be tested. This will allow a better definition of the optimal operating range. A third DoE, based on the results of prior DoEs is then fashioned to further refine results. CONCLUSIONS Quality function deployment (QFD) and design of experiments (DoE) are powerful tools that can be used to maximize information derived from a minimum number of tests, if used properly. Maximum sludge quality (highest solids) achievement at lowest cost can be found by these methods. Other factors can be added and tested in this manner as well, to reach optimal conditions.

REFERENCES Tague, Nancy R. (2005) The Quality Toolbox, 2nd ed.; ASQ Quality Press, Milwaukee, Wisconsin. Keller, Paul (2005) Six Sigma Demystified, A Self-Teaching Guide; McGraw-Hill, New York, New York. Water Environment Federation (1987) Operation and Maintenance of Sludge Dewatering Systems; Manual of Practice No. OM-8; Alexandria, Virginia.

Copyright 2008 Water Environment Federation. All Rights Reserved

756

Вам также может понравиться