Verification of automatic coagulant dosage control technology based on aluminum concentration at a water purification plant

Realization of rational maintenance is becoming more and more necessary in Japanese water purification plants as the number of available skilled plant operators decreases with the retirement of older workers and the amount of water use decreases with an aging society and overall decreasing population. Generally, in Japanese treatment plants, a coagulation-sedimentation process is controlled by feedback (FB) control based on settled water turbidity and feedforward (FF) control based on feed water quality. However FB control has a delay originating from the retention time of the sedimentation tank. Therefore, settled water turbidity may exceed a set value when the feed water quality changes rapidly due to heavy rains, and the experience and know-how of expert operators is required to implement manual control. Several groups have studied methods to deal with the sudden change of feed water quality (Annadurai et al. 2004; Lin et al. 2008). One method to increase coagulant dosage by predicting aggravation of feed water quality has been suggested. However, it tends to inject surplus coagulant. When using an aluminum (Al) component coagulant such as polyaluminum chloride (PACl), this means an increase of Al concentration in tap water which may be problematic since excess Al intake is suspected of causing chromaticity increase. Therefore a technique to shorten the delay time for FB control should be considered. For making an early judgment about coagulation dosage, a coagulant dosage control method using growth start time of flocs has been proposed (Yamaguchi et al. 2010). The electric potential value that is near the zeta electric potential can be measured by a streaming current detector (Dentel et al. 1989a, b; Maeda et al. 2005; Xia et al. 2007), or laser light scattering (Watanabe et al. 2008; Gregory 2009).

The authors have proposed improved FB control for coagulant dosage control (Yokoi et al. 2009). In particular, an indicator based on residual Al of water in the mixing tank was developed to shorten the delay time for FB control. Aluminum, mostly in the form of PACl, is commonly used as a coagulant in Japanese water purification plants. Residual Al is the sum of the concentrations of Al included in small flocs and dissolved in the mixing tank water. The Al residual rate is divided by the residual aluminum concentration in the coagulant dosage, and it is associated with the coagulation state by the jar test. A laboratory bench-scale coagulation-sedimentation process apparatus has been used to carry out a coagulation dosage control experiment with conditions that simulated rapid changes of feed water quality (Sangu et al. 2011). The effectiveness of the coagulant dosage control system based on the Al residual rate was verified.

In this paper we report the effects on the coagulant dosage control system based on Al residual rate applied to various river water quantities.

Figure 1 is a flow diagram of the new coagulant dosage control system. The developed system is mainly comprised of a coagulant feeding facility, a flocs classification system, an aluminum concentration measuring system, coagulant dosage operational equipment, and equipment for water analysis (turbidity, chromaticity, pH, alkalinity, water temperature). The coagulant dosage is decided from the FF and FB controls. In the developed system, the value PACl₀ for the FF control is calculated from two feed-water turbidity values. The value ΔPACl of the FB control is calculated from the difference between the Al residual rate and its target value. In a conventional system, the value of the FB control is calculated from the difference between settled water turbidity and its target value. The developed system differs from the conventional system regarding the sampling point of water for FB control and the water quality to be measured. These changes should allow the developed system to respond to rapid changes of feed water quality by shortening the delay time.

In this study, the developed system was implemented on a bench-scale coagulation-sedimentation process apparatus, and the apparatus was transferred to an actual water purification plant in December 2010. Figure 2 shows the specifications and setup of the coagulation-sedimentation process apparatus for the coagulant dosage control test. The main components were the mixing tank (capacity 10 L, residence time 5 min), the flocculator (120 L, 60 min), the sedimentation tank (280 L, 140 min), the flocs classification system, and the Al concentration measuring system. There were two lines (line 1 and 2); only line 1 had the flocs classification system and the Al concentration measuring system. The PACl coagulant was injected into the feed water in the mixing tank. In the flocculator, the water from the mixing tank was slowly stirred and floc growth was promoted. Then the water in the flocculator was supplied to the sedimentation tank. In the sedimentation tank, the flocs settled down and the settled water flowed to an outlet port. Test conditions of the coagulation-sedimentation process apparatus were set by referring to Japanese water service facility design indicators (JWWA 2000; JSCE 2004). The apparatus was operated automatically to control coagulant dosage from feed water turbidity, residual Al concentration, and settled water turbidity.

Part of the mixing tank water was supplied to the flocs classification system and the flocs were classified. When mixing water includes a large quantity of suspended matter and coagulant, Al concentration measurement is difficult. In addition, accuracy in judging excess coagulant and coagulant deficiency is improved by choosing small flocs which are the last to settle. Therefore, processing water classified as being rich in small flocs was supplied to the Al concentration measuring system and the residual Al concentration was measured. A cylindrical rotating strainer (December 2010–October 2011) and a sloped pipe (November 2011–September 2012) were used as the flocs classification system during the respective periods. Tables 1 and 2 summarize operating conditions of the cylindrical rotating strainer and the sloped pipe. When the coagulation processing is appropriate, the number of initial micro-particles decreases and particles in the feed water can be processed at high efficiency by filtration (Ebie et al. 2007). If coagulation processing is poor, there will be many small flocs (particle sizes less than 10 μm), so a 10 μm particle filter was used in this study.

The processing water was supplied to the Al concentration measuring system and residual Al concentration was measured. To measure residual Al concentration by absorption spectrophotometry, Eriochrome Cyanine R (ECR) reagent and acetic acid buffer solution were used. A flow cell type apparatus and a batch type apparatus were used as the Al concentration measuring system. The flow cell type apparatus was used from December 2010 to March 2012. The batch type apparatus was used after April 2012. Table 3 lists measurement conditions of the two sets of apparatus. A predetermined amount of the ECR regent (Merck, Ltd) was dissolved in pure water and the ECR solution pH was adjusted to <1.9 using hydrochloric acid. This pH was chosen to prevent deterioration of the ECR reagent and to dissolve Al included in small flocs. The ECR solution would be normally added to processing water after the acetic acid buffer solution addition. However, in this study, the ECR solution was added before the acetic acid buffer solution to dissolve Al included in small flocs. For the batch type apparatus the diluted Al standard solution (AlK(SO₄)₂, Wako Pure Chemical Industries, Ltd) was used, and measurement was possible with an error less than 0.02 mg/L (R² = 0.995) in an Al concentration range of 0–0.5 mg/L.

As the NaOH solution dosage control, FB control was carried out by pH in the mixing tank water and manual adjustment (the water purification plant reference) of the NaOH solution dosage infusion rate based on feed water turbidity (almost that of the coagulant dosage).

Table 4 gives experimental conditions during two periods of high feed-water turbidity. Case 2 (feed water turbidity > 100 mg-kaolin/L) occurred in September, as an effect of typhoon #15 of 2012.

The effectiveness of the developed system was evaluated on two points:

• Settled water turbidity: The difference between the target value and the actual value of settled water turbidity was compared with the same difference for the conventional system.

• Coagulant dosage: The corrected value of the developed system was compared with the value of the conventional system.

Figure 3 shows results of coagulant dosage control test case 1 in which the feed water turbidity changed during a low water temperature period (February). When feed water turbidity increased, the developed system adjusted the coagulant dosage immediately. Settled water turbidity of the conventional system increased to 0.9 mg-kaolin/L, above the target value of 0.5 mg-kaolin/L, but the developed system was able to keep the settled water turbidity lower than the target value.

Figure 4 shows results from the water purification plant during typhoon #15 (September). Figure 5 shows results of coagulant dosage control test case 2. At the water purification plant, feed water turbidity increased to 250 mg-kaolin/L and gradually decreased to the original turbidity after 4 days. When feed water turbidity reached the maximum, the coagulant dosage changed with less than 100 mg/L fluctuation, and coagulant dosage decreased afterwards. Settled water turbidity increased with a rise in feed water turbidity and reached 1.5 mg-kaolin/L. It gradually decreased, but was more than 1.0 mg-kaolin/L afterwards until feed water turbidity became less than 30 mg-kaolin/L. For the bench-scale apparatus, feed water turbidity increased from about 10:00 on September 9 and reached the upper limit level of the turbidity meter for feed water (270 mg-kaolin/L) at 19:00. Then, alkalinity decreased to 11 mg/L and gradually recovered afterwards. Conversely, pH decreased from 7.38 to 7.02. When feed water turbidity increased, residual Al concentration decreased from 0.23 to 0.13 mg/L. In addition, Al residual rate fell to 2.1%. This value was lower than the target value of the Al residual rate and it showed that excessive PACl was injected. Therefore the developed system carried out the adjustment to decrease coagulant dosage unlike the conventional system. The developed system used a dosage of 120 mg/L whereas the conventional system dosage was 170 mg/L when feed water turbidity was the maximum.

ΔTu_s (the difference in settled water turbidity before feed water turbidity change) of the conventional system continued decreasing and became –1.0 mg-kaolin/L from settled water turbidity before the test. In contrast, ΔTu_s of the developed system was less than approximately +0.5 mg-kaolin/L. The developed system was able to maintain settled water turbidity constantly in comparison with the conventional system. In addition, the developed system was able to reduce coagulant dosage 14% in comparison with the conventional system (the sludge removal period was excluded). This suggested that judgment of the excess or lack of coagulant is possible based on residual Al concentration.

Figure 6 shows settled water turbidity when applying the developed system. When feed water turbidity increased to 200 mg-kaolin/L, the settled water turbidity was maintained within ±0.5 mg-kaolin/L of the target value (0.5 mg-kaolin/L). Regarding the change speed of 30 mg-kaolin/h, which was the change speed of the natural feed water, settled water turbidity was maintained within ±0.5 mg-kaolin/L of the target value. In addition, when the change speed of feed water turbidity was more than 200 mg-kaolin/h (artificial feed water), the rise in settled water turbidity was controlled to less than 1.0 mg-kaolin/L. Therefore the developed system was able to support rapid changes in feed water turbidity.

The bench-scale coagulation-sedimentation process apparatus was installed in a Japanese water purification plant in December 2010, and the coagulant dosage control test using actual feed water was started. In this paper, the effect of the coagulant dosage control system based on aluminum residual rate was verified for various feed water qualities. The developed system was confirmed to be effective for maintenance of settled water turbidity and adequacy of coagulant dosage. For changes in actual feed water quality at the water purification plant over a period of more than one year, the developed system was able to maintain settled water turbidity at less than 1.0 mg-kaolin/L. Therefore it was judged that the developed system can be applied to actual apparatus in water purification plants.