Polyelectrolyte Flocculation of Waste Activated Sludge in Decanter Centrifuge Applications: Lab Evaluation by a Centrifugal Compaction Test

Sludge and conditioners

Fresh sludge samples were obtained from the clarifier underflow of the wastewater treatment plant of the Monsanto site in Antwerp (Belgium). Mixed liquor suspended solids and volatile fraction were determined according to the procedures in Standard Methods (APHA-AWWA-WEF, 2005) and the sludge volume index according to the Dutch norm NEN 6624 (1982). The centrifugal compaction tests with polymer-conditioned sludge were performed with sludge samples taken on three different days, with days 1 and 2 (as replicate) in the same week, which secured almost the same sludge characteristics. Another test on day 3 was performed with the sample taken 2 months later when the calcium concentration was significantly lower (Table 1). The additional comparisons between the polymers, by means of the sheared CST test and the measurement of the floc size distributions (FSD), were performed at a later time and hence on different sludge.

The three polymers P1, P2, and P3 (Table 2) were obtained as 50 wt% active substance (a.s.) from the suppliers in emulsion form. They were first diluted to a concentration of 0.66 wt% a.s. and stirred gently during 1 h. Next, these aged polymer solutions were further diluted to 0.20 wt% a.s. and homogenized during 15 min using a shaking table at 150 rpm (Shaker VKS 75 Control; Bühler) to finally obtain the same active polymer concentration as that of the polymer solution being applied in the field to condition the sludge feed in the decanter centrifuge. Polymer P1 is a tailor-made blend of linear and cross-linked cationic polymers with a low degree of cross-linking as reported by the supplier, and is currently used at Monsanto's centrifuge-dryer installation for >10 years. The alternative polymers P2 and P3 are copolymers of acrylamide and quaternized N,N-dimethylaminoethyl acrylate. Charge densities (CD) and MW of the poly-electrolytes, as provided by the suppliers, are listed in Table 2. Polymer P2 has a higher degree of cross-linkage than polymer P1, and polymer P3 has additionally a higher CD on top of the higher degree of cross-linkage. According to the review article of Bolto and Gregory (2007), the polymers can be regarded as having a high MW (>10⁺⁶) and high CD (50–100 mol% of the cationic component). Clay additive (Sepiolita-100; Tolsa), another sludge feed conditioner used at this particular centrifuge-dryer installation, was also taken from the installation in the field.

Sludge conditioning

The sludge conditioning applied in the lab was aimed at obtaining similar sludge characteristics during the test as during normal operating conditions, implying the addition of clay and cationic polymers (Peeters et al., 2009b). More specifically, while gently mixing 800 mL of sludge with a lab magnetic stirrer, 1.7 g of clay was added, yielding a clay-to-biosolids ratio of ∼0.1 kg clay/kg biosolids. After homogenization of this mixture during 2 min, polymer (0.20 wt% a.s.) was added, equivalent with volumes of polymer used in the field from 15 to 75 L polymer (as 0.66 wt% a.s.) per m³ sludge to investigate the effect on the sludge dewaterability. Next, mechanical stress was applied on the flocs in a typical Jar test apparatus (Floc-Tester, Lovibond ET750) to impose shear forces, thus simulating shear forces acting on the flocs in the decanter centrifuge: the conditioned mixture was agitated for 15 min in a 1 L beaker with a flat single-bladed paddle type stirrer at a controlled speed (shear) of 230 rev/min. The applied shear in coagulation/flocculation processes is usually characterized by the average velocity gradient G [s⁻¹] (Ivens, 2000): \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} G = \sqrt {\frac {P} {V \times \mu}} \end{align*} \end{document}

where P [W] is the power transferred from the stirrer to the suspension; V [m³] is the volume of the stirred suspension and μ [Pa·s] is the dynamic viscosity (0.001 Pa·s). The power P is given by (Geankoplis, 1993): \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} P = P_0 \times N^3 \times D^5 \times \rho \end{align*} \end{document}

where P₀ [/] is the impeller power number; N is the impeller speed [rev/s]; D is the impeller diameter [m], and ρ [kg/m³] is the density (1,000 kg/m³). Taking into account the impeller power number of 0.3 and diameter of 7.5 cm, the estimated G amounts to 220 s⁻¹.

Müller and Dentel (2002) have shown that the mixing conditions during the addition of polyelectrolyte have a particular influence on the sludge dewatering behavior. Polymer dose, mixing intensity, and mixing time should match to ensure that optimal dewatering results are obtained. Compared to linear polymers, to get good dewatering results, higher quantities of cross-linked polymers need to be added, in combination with an extended mixing time to get the best dewatering performance. However, once the best dewaterability is obtained (at a certain dosage-mixing combination) for a cross-linked polymer, this optimal dewaterability is sustained longer when shearing continues until finally the dewatering deteriorates (Müller and Dentel, 2002). In this study, the mixing intensity (estimated G of 220 s⁻¹) and time (15 min) was always the same for the different polymers and dosages before the sludge was centrifugal dewatered in the lab, as is the case in the industrial installation. The separate effect of higher mixing intensities on the suspensions was evaluated with the sheared CST test and the FSD measurements.

Centrifugation

The conditioned sludge was allowed to settle for 5 min in the 1 L beaker, and the supernatant was decanted. Then, spin tubes were filled with 50 g of pre-thickened solids (dryness typically ranging from 10% to 15% dry substance [DS]). Next, the solids were exposed to g-forces (horizontal head) for a long time (45 min) in a lab centrifuge (4,400 rev/min; radius of rotation R being 135 mm at the bottom of the tubes) after which the compacted cake had reached equilibrium with all free water expressed out of the cake. After this centrifugation step, the liquor above the cake surface was decanted and separate cake layers of ∼5 g were carefully removed for dryness analysis (DS_i); the exact mass of the cake segments removed (w_i) was recorded. Also, the cake heights were recorded before (h_i) and after (h_i+1) the cake segments were removed from the spin tubes (Fig. 1). With these data of the stratified cake, the compaction curves were obtained by plotting (log₁₀ of) cake dryness of the different segments versus the respective (log₁₀ of) calculated compressive pressures acting on the center of the segments, following Equation (7) (see further in the Section Centrifugal Compaction ).

OPEN IN VIEWER

FIG. 1.

Spin tube filled with compacted sludge.

Sheared CST test

The sheared CST test is being used to evaluate chemical pretreatment (flocculation) in case of a centrifuge application (Moody and Norman, 2005) to simulate the very high levels of shear the flocculated substrates encounter during their turbulent entry in the decanter centrifuge (Leung, 1998; Records and Sutherland, 2001). The method consists of adding the pretreatment chemical to be tested to the feed suspension and subjecting the suspension to various levels of shear, achieved by high-speed stirring, after which the dewaterability of the suspension is assessed by measurement of CST. The relative floc strength is assessed by plotting CST values versus time of stirring. A feed suspension with strong flocs will show only gradual increase in CST with stirring time, whereas one with weak flocs will show a significant increase in CST with stirring time. The CST of the conditioned sludge, already sheared during the sludge conditioning as described in the Sludge conditioning section, was measured after additional 0, 10, 20, 30, 40, 100, and 160 s of shearing at the high speed of 1,000 rev/min (pitched blade turbine impeller with a power number of 1.0 and a diameter of 6 cm implying an estimated G value of 2,100 s⁻¹) in the beaker as shown in Fig. 2. CST measurements were 3 times repeated. According to Moody and Norman (2005), the greater intensity of mixing (speeds even up to 2,000 rev/min) provided by the laboratory stirrer probably provides better simulation of the high shear conditions inside a centrifuge.

OPEN IN VIEWER

FIG. 2.

Sheared capillary suction time (CST) setup.

FSD measurement

Samples of conditioned sludge as described in the Sludge conditioning section were transferred to a particle sizer (LS 230; Beckman Coulter) for measuring the FSD in the range from 0.04 to 2,000 μm using light scattering technique. Sample volumes were added to obtain a light obscuration of ∼20% for reliable measurements. FSDs were measured during 29 min while being circulated at a flow of 27 mL/s through the plastic tubing with an internal diameter of 5 mm and a length of 1.1 m. The average velocity gradient G was calculated in this case based on the power dissipated along the tube, with the loss of pressure head in turbulent regime given by the Darcy-Weisbach equation (Ivens, 2000). With a Fanning pipe friction factor of 0.01 the estimated G value is equal to 3,340 s⁻¹. The bulk mixture in the tank (∼0.85 L) is stirred with an agitator at 670 rev/min (propeller impeller with a power number of 0.2 and a diameter of 4 cm inducing an estimated G value of 200 s⁻¹).

To summarize the obtained FSDs, the D[50] (median floc size) and the interquartile range [IQR] (the difference between the 75th and 25th percentile, D[75]−D[25]) were used as measures for respectively the average floc size and its spread (Dytham, 2003). The colloidal fraction of the FSDs was defined as the total volume fraction of all flocs with a size <10 μm, which has previously been used by Feitz et al. (2001) to study the effect of floc size and structure on centrifugal dewatering.

Full-scale experiments

After the lab evaluations, the performance of the polymer P3 was evaluated in the industrial centrifuge-dryer installation of the Monsanto Antwerp WWTP (Peeters, 2010b). For this, polymer P3 was used in two campaigns of, respectively, 11 and 14 days, changing the polymer dosing from 23 to as low as 13 kg a.s./ton DS. Before, in between and after these two campaigns, the (standard) polymer P1 was used during, respectively, 9, 10, and 14 days. The first reduction in polymer dosing below 17 kg a.s./ton DS (the critical polymer dosing in terms of cake dryness after centrifugal dewatering in the lab) was evaluated with the best performing polymer P3 on lab scale. Because of the steep decline in final product dryness at the reduced dosing of polymer P3 (which performs better than polymer P1 in the lab), it was decided to keep the polymer P1 dosing always above 17 kg a.s./ton DS. Daily samples were taken from the sludge feed to the centrifuge-dryer system and analyzed for the mixed liquor suspended solid concentration, and the dryness of the final dried product was determined. It must be mentioned that sampling of the cake leaving the decanter centrifuge (for analysis of the dryness) is, unfortunately, not possible because the centrifuge-dryer system is one compact, enclosed machine (Peeters, 2010b). We, therefore, have to rely on the dryness of the finally dried product as a measure for the field performance. Industrial practice at Monsanto over the past years has, however, shown that the performance derived on the basis of a (compared to the currently proposed procedure) somewhat simplified spin tube (calculation) procedure always exhibited a clear match with the obtained final dryness and, hence, with the compaction performance in the decanter centrifuge-part of the system (see also Peeters et al., 2009a; Peeters, 2010b). Details on this particular centrifuge-dryer system can be found in earlier published work (Peeters, 2010b).

Statistics

The 95% confidential intervals were obtained with the MINITAB^® software package.

Centrifugal compaction

In circular motion the centrifugal force F_c acting on a particle of mass m [kg], at a radial distance R [m] from the center of rotation, is given by: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} F_c = m \times R \times \omega^2 \tag{1} \end{align*} \end{document}

with ω the angular speed [rad/s]. When submerged in water, the water buoyancy reduces the weight of the solid particle under centrifugal gravity and as a consequence the centrifugal force equals: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} F_c = V_s \times \mathit \Delta \rho \times R \times \omega^2 \tag{2} \end{align*} \end{document}

where V_s is the particle volume [m³] and Δρ=(ρ_s–ρ_w) is the difference in density of the solid particle and water, respectively [kg/m³].

Figure 1 depicts schematically a spin tube filled with activated sludge. The sludge solids in a layer of thickness dR at a radial distance R_i from the axis of rotation exert a downward force that is given by: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} dF_c = dV_S \times \mathit \Delta \rho \times R_i \times \omega^2 \tag{3} \end{align*} \end{document}

where dV_s is the solids volume in the sludge layer [m³], or \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} dF_c = dm_S \times \frac {\mathit \Delta \rho} {\rho_S} \times R_i \times \omega^2 \tag {4} \end{align*} \end{document}

with dm_s being the solids mass [kg] in the cake layer of thickness dR at radius R_i. At each depth of the sludge cake, the total solids stress experienced by the solids is determined by the cumulative solids stress generated by the sludge layers higher in the cake. The solids stress P_s, or compaction stress, exerted on the solids at radius R_i in Fig. 1 is equal to the sum of the contributions from all the layers higher in the cake and is determined from: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} P_S (R_i) = \int \limits_1^i dP_S = \int \limits_1^i \frac {dF_C} {A} = \frac {\mathit \Delta \rho} {\rho_S} \times \frac {\omega^2} {A} \times \int \limits_1^i R_i \ dm_i \tag {5} \end{align*} \end{document}

The symbol A represents the surface area of the layer, that is, from the spin tube [m²].

Approximate determination of solids stress

Consider the right side of Fig. 1 where the sludge cake is divided in different segments to determine approximately the solids stress exerted on the center of every segment in the sludge cake by a sum. Starting from equation (5), the solid stress exerted on the solids in the center of segment i at radius R_i is approximately equal to the following sum: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} P_{s} (R_{i}) \cong \frac {{\frac {\Delta \rho} {\rho_{s}}} \times \omega^2 \times [m_1 \times R_1 + m_2 \times R_2 + \cdots + m_{i - 1} \times R_{i - 1} + {\frac {m_i} {2}} \times R_{i}]} {A} \tag{6} \end{align*} \end{document}

Finally, based on the parameters as shown on the right side of Fig. 1, the solids stress (in kPa) executed on a segment i is found with following formula: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*} P_{S,segment \ i} \cong \frac{\left(1-\frac{\rho_w}{\rho_s}\right) \times \left(2\pi{\times}\frac{N}{60}\right)^2 \times \left[\sum_{j = 0}^{j = i-1}\left[\left(w_j \times \frac{\% DS_j} {100}\right) \times \left(R - \frac{h_j + h_{j+1}}{2}\right) \right] + \left(\frac{w_i}{2} \times \frac{\% DS_i}{100}\right) \times \left(R - \frac{h_i + h_{i+1}}{2}\right)\right]} {\frac{\pi D^2_{tube}}{4} \times 1000} \tag {7} \end{align*} \end{document}

where N is the speed of rotation [rpm], w_i is the mass of cake layer i [kg] and %DS_i is the cake dryness of layer i (%), R is the distance from the bottom of the tube to the axis of rotation [0.135 m], h_i is the height of the segments relative to the bottom of the spin tube (see Fig. 1), and D_tube is the internal diameter of the spin tube [0.028 m].

Centrifugal compaction test

Figure 3 presents the 36 compaction curves obtained. The curves obtained with the samples D1 and D2 (replicate) are shown with the same symbol to keep the figure readable. The dissection of the five cake layers, obtained after centrifugation, is clearly seen by the different clusters of data. As a next step, cake dryness was determined at 20 kPa compaction stress by interpolation on the individual compaction curves. The results are shown in Fig. 4. For the samples D1 and D2, evaluations were performed in a wide range from ∼5 to 30 kg polymer a.s./ton DS, including replicates in the middle of this range (at an applied dosing of 45 L polymer/m³ sludge corresponding with ∼17 kg polymer a.s./ton DS). The replicates show that the results from the centrifugal compaction test are repeatable, confirming earlier results, obtained with a somewhat simplified calculation of the compaction stress (Peeters et al., 2009b). For all polymers, the cake dryness increased significantly with increasing polymer dosing, up to ∼17 kg a.s./ton DS. Further increase of polymer yielded comparable cake dryness on lab scale, once this plateau of maximum achievable cake dryness was obtained, with even a small deterioration in cake dryness at the highest dosing. Figure 4 clearly shows that the medium cross-linked polymer P3 with the highest CD resulted in a higher achievable cake dryness, followed by the other medium cross-linked polymer P2, whereas the polymer P1 with the low degree of cross-linkage yielded the lowest cake dryness. The same results, that is to say, the relative performance of the three polymers, were obtained 2 months later with sample D3, with the two medium cross-linked polymers P3 and P2 outperforming the low degree cross-linked polymer P1, though the cake dryness was shifted downward consistently with ∼2%DS. The difference in cake compactibility can be explained by the low Ca²⁺ concentration in the wastewater in the case of sample D3 (∼100 ppm) compared to the samples D1 and D2 (∼800 ppm). The effect of Ca²⁺ in the water surrounding the sludge flocs on its dewaterability was shown in the past by CST measurements on sludge samples of this WWTP (Peeters and Herman, 2007) and is described in literature (Higgins et al., 2004; Nguyen et al., 2008).

OPEN IN VIEWER

FIG. 3.

Compaction curves obtained with samples D1 and D2 (replicate) (shown with the same symbol) and D3 for three polymers P1, P2, and P3 at different dosings: 15, 30, 45, 60, and 75 L/m³.

OPEN IN VIEWER

FIG. 4.

Cake dryness at a compaction stress of 20 kPa as a function of the amount of administered active substance of polymer per ton dry solids.

Sheared CST

In Fig. 5 the sheared CST measurements (246 in total) are shown at different polymer dosings, and after various stirring times of the samples ranging from 0 to 160 s, for polymer P1 and P3. When polymer was added to the raw sludge, larger flocs were formed, and some initial pronounced decrease in CST compared to the raw sludge CST was observed. Sludge conditioned with a low amount of 4 L/m³ (1.2 kg a.s./ton DS) yielded an improved filterability (CST of 10 s) compared to unconditioned sludge (25 s), but the flocs formed at these low dosages broke up, which is shown by a slight increase of CST at increased stirring times. At the dosages of 9 and 19 L/m³ (2.6 and 5.6 kg a.s./ton DS), the CST remained stable at 6 s. However, upon increase of the polymer dosage, CST values started to rise again and became larger than the raw sludge samples' CST values, which indicates a deterioration in dewaterability. This is in contradiction to the clear improvement of the sludge dewaterability at higher polyelectrolyte doses starting from 17 kg a.s./ton DS, demonstrated with the centrifugal compaction tests (Fig. 4). These data demonstrate that the CST is an unreliable parameter to assess biological sludge dewaterability on lab scale in the case of sludge dewatering applications with decanter centrifuges where dewatering takes place mainly by expression rather than by filtration. According to Christensen and coworkers (1993) the unreliability of the CST may be ascribed to the increased viscosity of the liquid phase when polymer overdosing is applied. At these high dosages, breakage of flocs by stirring leads to further consumption of polymer in solution (Langer et al., 1994), reducing its viscosity and yielding a lower CST value at longer shearing times.

OPEN IN VIEWER

FIG. 5.

Sheared CST measurements for polymer P1 and P3 (three repeats of the CST for every combination) (95% confidential interval).

A remarkable difference was observed between both polymers at the high dosages. At 37 L/m³ (11 kg a.s./ton DS) polymer P1 showed a higher overdosage (higher CST) than polymer P3, but at 75 L/m³ (22 kg a.s./ton DS), polymer P3 showed a higher overdosage than P1. Even after 160 s of stirring excess (unconsumed) polymer remained in the liquid phase, which indicates that less polymer was consumed for reflocculation, whereas in the case of P1 all polymer was depleted from solution. At the highest dosage of 75 L/m³ the conditioned sludge flocs were more shear resistant in the case of polymer P3 with a higher degree of cross-linking and a higher CD, shown by a smaller slope of decreasing CST versus stirring time, and a surplus of unconsumed polymer remained in solution. Compared to polyelectrolytes without or a lower degree of cross-linkage, the modified (highly branched and structured) polymers require a greater dosage to reach its optimum performance, but the flocs are larger and more shear-resistant (Dentel et al., 2000; Dentel, 2001; Müller and Dentel, 2002; SNF Floerger, 2003). Additionally, in the review of Bolto and Gregory (2007) it is found that polyelectrolytes of higher CD tend to be more effective, which is (in addition to the higher degree of cross-linkage) possibly part of the explanation for the better performance of polymer P3.

Floc size distribution

It is recognized that particle size has a significant effect on the solid/liquid separation behavior of a suspension. For instance, for filtration to work efficiently, the target properties for the particles would be to have as large a size as possible, and have a monosize distribution (Wakeman, 2007b). Moreover, the finer particles in the particle distribution are important for characterizing a dewatering process with a filter as well as with a centrifuge (Feitz et al., 2001; Wakeman, 2007b; Sanin et al., 2011). An improved dewatering is shown to correspond with a smaller proportion of these fine particles [e.g., an increase in sludge cake dryness after centrifugation from 16% to 20% DS was correlated by Feitz et al. (2001) with a decrease of the fines from 30% to 24%]. Besides, the past (shearing) history of a sludge is of considerable importance for the shear resistance of the finally flocculated suspensions, and the optimum past (shearing) conditions in terms of the final shear resistance varies depending on the polymer type (Müller and Dentel, 2002). However, in this study, as aforesaid, mixing intensities and duration were fixed reflecting the constant operational conditions of the industrial installation.

The median floc size for three polymer dosages (45, 60, and 75 L/m³) as function of the circulation (shearing) time in the Coulter is presented in Fig. 6 for polymers P1 and P3. Unconditioned sludge had a median floc size of 37 μm, which remained stable during the FSD measurements. For the polymer-conditioned sludge, throughout the measurements, the median floc size decreased, indicating that the shear exerted by the instrument during measurement process caused floc breakage. It was found that at the lowest dosage of 45 L/m³ (16.1 kg a.s./ton DS), the floc size was larger in the case of the polymer P1, but at the highest doses of 60 and 75 L/m³ (corresponding to 21.5 and 26.9 kg a.s./ton DS, respectively), polymer P3 yielded larger flocs, also at the end of the measurement process.

OPEN IN VIEWER

FIG. 6.

Median floc size (D[50]) for the polymer P1 and P3, at three polymer dosings (45, 60, and 75 L/m³), as function of circulation (shear) time in the Coulter device.

At the left side of Fig. 7, the IQR is presented as a function of the D[50]. The overall effect of an increased shear applied on the flocculated suspensions was a reduction in the D[50] (as already shown in Fig. 6) and a reduction in the IQR, which indicates that the distributions become smaller; that is to say, a more uniform distribution resulted as result of the shear forces. There is one remarkable difference between polymer P1 and polymer P3 at the higher dosages. In the case of polymer P1, the bigger flocs obtained by increasing the dosing from 60 to 75 L/m³ had a larger IQR at a certain median floc size (in Fig. 7, the curve is shifted to the left), whereas in the case of polymer P3, the bigger flocs obtained at a dosing of 75 L/m³ had a smaller IQR at a certain median floc size than for a dosing of 60 L/m³ (in Fig. 7, the curve is shifted to the right). It appears that the more cross-linked and more densely charged polymer P3 yielded a more uniform FSD at the highest dosing, in addition to a bigger floc size.

OPEN IN VIEWER

FIG. 7.

Interquartile range (IQR, D[75]−D[25]) and colloidal fraction (total vol% of flocs <10 μm) as function of the median floc size (D[50]). Effect of an increased circulation (shear) time (from 0 to 29 min) is indicated with an arrow.

As shown at the right side of Fig. 7, the overall effect of shear was the increase of the particle fines content along with a decrease in the D[50]. The colloidal fraction was at a very low level of 0.3 vol% at the beginning of the FSD measurements in case of the highest dosages of 60 and 75 L/m³ for both polymer P1 and P3, and increased to a level of 1 vol% after 29 min. At the lower dosage of 45 L/m³, the initial proportion of fines increased as result of the shearing from 0.2 to 1.8 vol% for polymer P1, and from 0.7 to 1.9 vol% for polymer P3. The colloidal fractions in this study are at such a low level that no significant effect on the centrifugal dewaterability can be expected. Besides, at the highest dosages, no difference at all existed between both polymers.

Conclusively, it is known that a greater dosage of the modified (highly branched and structured) polymers is required to reach optimum performance, but the flocs attained are larger and more shear-resistant (Dentel et al., 2000; Dentel, 2001; Müller and Dentel, 2002; SNF Floerger, 2003). The higher CD of polymer P3 serves as an additional explanation for its better performance (Bolto and Gregory, 2007).

Compatibility between centrifugal compaction test and full scale conditions

The dryness of the final dried product is shown in Fig. 8 as function of the polymer conditioning applied to the decanter centrifuge. As aforementioned, cake sampling after the mechanical dewatering step is not possible for this particular centrifuge-dryer system, and hence the dryness of the sludge particles after the additional thermal drying stage was taken to evaluate the polymers' performance in the field. The dosages lower than 17 kg a.s./ton DS were only tested in the field for the best performing polymer P3 on lab-scale.

OPEN IN VIEWER

FIG. 8.

Final dried product of the industrial centrifuge-dryer system.

A relatively high spread in the final dried product dryness was observed, for example, for the polymer P1 at dosages higher than 20 kg a.s./ton DS the dryness amounted to 85.0%±5.4%DS (average±standard deviation). Though, this variation should not be surprising taking into account the variation already induced by the sludge decanter centrifuge itself. The reference Decanter Centrifuge Handbook (Records and Sutherland, 2001) mentions that for a decanter centrifuge, at best, a cloud of points will be seen on a graph when evaluating the effect of different polymer dosages on the sludge cake dryness. Because of the low differential speed in the range from 1 to 4 rev/min (Peeters, 2010a) to transport the sludge cake out of the centrifuge bowl, it takes a long time to arrive at steady-state conditions and, in practice, the process never reaches steady state and is always under a transient condition (Leung, 1998). As a result of the inevitable variation, a higher (but not statistically significant) final dryness was observed in the case of polymer P3 (Fig. 8). A longer evaluation period and hence more data would be needed here to be more conclusive.

However, the data gathered from the field test with the polymer P3 (Fig. 8) are in very good agreement with the data from the centrifugal compaction test (Fig. 4): a steep decline in final product dryness was observed when a dose below 17 kg a.s./ton DS was applied. At that moment, the dryness of the cake leaving the decanter centrifuge, immediately entering the initial stage of the flash dryer, drops significantly, which cannot be compensated by the dryer capacity, yielding the significantly lower final dryness. Also a clear deterioration in centrate quality was observed, accompanied by an increased vibrational level of the decanter centrifuge (results not shown), which was resolved by increasing the polymer dosing again. It must be (re-) mentioned that in case the practical polymer dosing would have been deduced from the “optimal” dewaterability in terms of lab CST measurements, serious operational problems would certainly have been the result. In other words, the CST measurements failed to provide appropriate data in terms of indicating achievable cake dryness after centrifugal dewaterability.

This evaluation shows the compatibility of the centrifugal compaction test with the dewatering taking place in the high-solids decanter centrifuge, including the sludge conditioning/shearing protocol, which is known to have an important effect on the subsequent dewatering process (Langer et al., 1994; Müller and Dentel, 2002), but which is at the same time the most difficult (unknown) part of lab simulations. According to Papavasilopoulos and Markantonatos (2001), the failure of laboratory tests to map behavioral aspects of polymer conditioning at full scale is generally attributed to differences in shear intensities applied in the laboratory and full scale. If the lab and full scale tests, however, do agree, then the lab procedure is likely to be generally predictive of the full-scale outcome (Dentel, 2010).