Octopus Technical Paper Presented at WEFTEC 2007

ACCURACY AND COST SAVINGS FROM AUTOMATIC OPTIMIZATION OF DEWATERING

Marty Davidson Alfa Laval

955 Mearns Road Warminster, PA 18974

ABSTRACT Improvements in sensor and automation technology in recent years have made it possible to design a system to control the key variable parameters in and around the dewatering process. By continuously analyzing feed conditions and output from the process, an automated optimization system can make necessary proactive adjustments to the internal settings of the dewatering equipment, the polymer dosage and the feed. The result is continuous process optimization, 24 hours a day, 7 days a week, with drastically reduced operator involvement. In late 2005, Alfa Laval tested automatic optimization at the City of San Diego Metro Biosolids Center (MBC). The optimization system provided “real-time” measurement of suspended solids content of the incoming sludge, suspended solids content in the centrate, polymer dosing rate and differential speed and conveyor torque of the decanter. These parameters are used in combination with the operational cost structure, factoring in MBC’s costs of polymer, disposal, and power. The system’s algorithms automatically vary the polymer dose, the differential speed of the decanter and the feed rate of sludge to ensure the optimal operation and minimize the overall cost of operation.

KEYWORDS Automation, optimization, dewatering, biosolids, centrifuge. INTRODUCTION The City of San Diego Metro Biosolids Center (MBC) is an essential component of the region’s 180 million gallon per day wastewater treatment. The facility serves to thicken and digest a combination of primary and secondary sludge, then dewater the digested sludge. MBC staff had been searching for ways to improve efficiency in the plant as part of a continual improvement process.

After being introduced to the concept of an automated dewatering optimization, MBC staff agreed to test the system at their facility. The test was designed to demonstrate whether such a system could achieve consistent peak performance from the dewatering process and thus reduce operational costs.

GOALS AND OBJECTIVES The primary goal of the study was to evaluate if automatic optimization of the dewatering process would cut costs and improve efficiency. The objectives were:

• To verify whether the automated optimization system could accurately measure process variables such as sludge feed solids, centrate solids and cake dryness.

• To verify whether the automated optimization system could reduce overall operating expenses through one or all of the following: increased cake dryness, reduced polymer consumption and/or reduced recycled solids.

BACKGROUND INFORMATION Sludge dewatering is one of the single most significant expenses incurred in most water treatment processes. In some cases, dewatering can account for nearly 30 percent of the total cost of treatment. This cost is slated to rise to an even higher percentage in the coming years driven by new environmental regulations and increasing personnel and transportation/disposal costs. Thus, attaining peak performance of the dewatering process is more important than ever today. Both suppliers and users of dewatering equipment have undertaken research aimed at achieving higher throughputs, reducing polymer consumption, and attaining cleaner centrate and dryer cake. The ultimate goal is to reduce the total cost per dry ton, the standard measurement of process efficiency. While many research projects have been underway, to date, one valuable area had not been examined in full – how to consistently optimize today’s increasingly sophisticated dewatering equipment. Under traditional operation, the continuously changing nature of the feed makes it difficult to optimize the dewatering process without intensive operator involvement. Typically, operators must continually monitor conditions such as variations in sludge concentration, consistency and quality. Then, they make manual adjustments to the process to compensate for these variations. In many cases, to be most effective process variations would actually require results of lab analysis; however, by definition, if lab results are available, it means it is already too late to take action on the findings. Because of this fact, even the most experienced operators cannot achieve continuous process optimization. Therefore, even with the best operators, the inevitable result is a trade-off between a very labor-intensive process and a sub-optimal process. Running at a sub-optimal level of performance has far reaching consequences, including lower dryness of the final cake (which will usually result in higher final disposal costs), higher polymer consumption, and more solids being re-circulated into the process through dirtier centrate.

In contrast, automated optimization ultimately results in a significant improvement in polymer consumption and/or dryness of the cake, a lower recycle load and improved operator peace of mind.

METHODOLOGY

For the test at MBC, two decanters were equipped with automated optimization systems: Decanter 5 (DC5) and Decanter 7 (DC7). A test was performed from November 2 through December 8, 2005. During this period, one decanter was operated and controlled manually, with adjustments being made by the plant’s staff. Simultaneously, Octopus by Alfa Laval automatically controlled and optimized the second decanter. The Alfa Laval Octopus system at San Diego Metro Biosolids Center consisted of:

• A stand-alone panel featuring an industrial PC used to run the system’s proprietary software. The software controls all communication parameters and optimization algorithms.

• Two sensors to monitor the suspended solids in the sludge feed and the centrate. • A specially developed tank that conditions the centrate and also houses the centrate

suspended solids sensor. • All necessary cables and other accessories.

The system oversaw the dewatering process on Decanter 7, making adjustments as needed to minimize its overall cost, measured as the cost per ton dry sludge of feed processed. Then MBC selected parameter limits based on the site’s procedures and priority. For example, MBC had the option to choose the minimum and maximum cake dryness to meet local operational constraints, such as their cake pumping capacity or disposal contract obligations. Alternatively, a minimum solids recovery level could be set to ensure that the recycle load was kept within acceptable limits. This allows the site’s priorities to be taken into account. To achieve optimization, the system monitors and controls the following process parameters:

• Suspended solids content and flow rate of the incoming sludge. • Suspended solids content in the centrate. • Polymer dosing rate. • Differential speed and conveyor torque of the decanter.

The optimization algorithms used these parameters in combination with the operational cost structure, factoring in the costs of polymer, disposal, power and water. The algorithms automatically varied the polymer dose, the differential speed of the decanter and the feed rate of sludge to ensure the optimal operation, resulting in minimization of the overall cost of operation. SYSTEM COMPONENTS Four individual system components particularly impacted operational performance:

• Hardware • Communication • Modes of operation • Optimization algorithms

Hardware To control the process effectively, the system needed accurate, reliable measurements of the suspended solids content in both the feed and the centrate. To achieve such accuracy, Alfa Laval used Hach Solitax® sensors. See Figure 1. These sensors provided measurement of both turbidity and color-independent measurement of solids content. By scattering infrared light into the target stream, they could measure the reflected light via two light sensors. One sensor measured turbidity, while both measured solid content. This infrared-duo scattered light method measured turbidity in the range of 0.001- 4000 NTU, and solids content from 0.001 to 150g/l. The Hach system also provided easy, one-point calibration and had a built-in wiper to clean the lens at preset intervals, ensuring reliable results. Figure 1 - Sensor

The feed sensor was installed in the sludge line using a special supplied mounting device. Another important hardware feature was the unique tank in the Alfa Laval Octopus system that helped to reliably measure the suspended solids content in the centrate. See Figure 2. This tank was specially designed by Alfa Laval to provide more precise measurements with the following features:

• The centrate flow was de-aerated to ensure that air bubbles did not distort the turbidity measurement, providing more accurate measurements.

• The positioning of the sensor and the automatic flushing routine prevented solids build-up in the vessel, which would otherwise eventually distort the sensor readings.

• The system periodically dosed diluted acid onto the lens surface to ensure that there was no build-up of scaling (e.g. struvite), keeping the lens surface as clean as possible and preventing distortion of the sensor’s measurement.

• The tank and the centrate pipe that supplied the feed to the tank were automatically flushed at regular intervals, enhancing measurement accuracy.

Figure 2 – Centrate Tank

The software itself was delivered installed in an industrial PC mounted in a NEMA 4X control panel. This panel was pre-fitted with all the necessary communication components as well as a modem that allowed remote access to update software or provide support services as needed.

Communication The system was designed to provide flexible communication to existing plant equipment such as the polymer pump, feed pump and decanter controller, and to other equipment such as the sensors and centrate tank. An OPC server provided communication with local control systems as needed. Depending on the user’s requirements, this communication can be set up in three different manners:

• Hard-wired, where all instruments are connected directly to the Octopus panel, bypassing any local control system (e.g. PLC or SCADA).

• Partially hard-wired, where existing plant equipment (feed and polymer pumps) is controlled locally while the sensors and centrate tank are hard-wired to the Alfa Laval Octopus panel.

• Fully soft-wired, where all equipment is connected to and controlled by the local control system.

A safety watchdog function is built into all communication routines. If communication with Octopus was interrupted at any time, process control automatically reverted to the local control system. Modes of Operation The system was able to run in one of four modes:

• In monitoring mode, control of the dewatering operation remained with the plant personnel using their own existing control system. Alfa Laval Octopus merely logged all relevant process parameters. This mode can also be described as standby mode.

• In set-point mode, Alfa Laval Octopus took control of the process, but the user manually

entered the set points for the key parameters, which the system maintained.

• In optimization mode, Alfa Laval Octopus took over full control of set points and automatically set values for feed rate, polymer dosing and differential speed. These values were based on real-time process analysis performed by the system’s optimization algorithms, within process limits set by the operators.

• In sequencer mode, used during start-up, the process ran for a preset length of time in set-

point mode before switching to optimization mode. The system had two levels of configurable user access, and the user interface differed based on the level of access selected. The operator level showed basic process values, while the

administrator mode showed both process values and real-time costs. Units and currency used were configured to local requirements. The following illustrates two screen captures from the system. Figure 3 depicts the standard user interface and Figure 4 shows the trend display for a 24-hour period of process development. Figure 3 – Standard User Interface

- Trend Display

Optimization Algorithms The optimization algorithms are advanced software routines specially developed by Alfa Laval. They incorporate Alfa Laval’s considerable process and decanter knowledge from years of industry experience. The algorithms’ operation can be likened to the actions that would be taken by a highly experienced operator based on a series of events. Continuous availability and early reaction to process changes avoids process deterioration during a delay where action is waiting to be taken.

Manual operation Octopus optimization

It also means that only small adjustments are required to compensate for process variations. The result is an overall smoother-running, more consistent process. Under stable operating conditions, Alfa Laval Octopus always seeks to reduce the overall dewatering cost by saving polymer, increasing the dryness of the final cake, or both, depending on the cost structure at the plant. For example, increasing the polymer dosage can enable the system to deliver drier cake and cleaner centrate, but this will be at the cost of additional chemicals. Alfa Laval Octopus analyzes the relevant costs and benefits, and takes the action necessary to ensure the greatest possible overall savings. Any actions are always subject to the limitations set by the user. For example, should the plant have a contract for disposal of sludge which stipulates a minimum cake dryness, Octopus will work within that limit, even though it may be cheaper to sacrifice dryness in exchange for savings in polymer consumption. If process conditions change and overstep the standard limitations, Octopus will take actions to bring the process back within acceptable limits. For example, a sudden increase in feed solids may result in a deterioration of centrate quality beyond the acceptable limits. The system takes immediate corrective action, by increasing polymer and/or by increasing differential speed. Octopus evaluates the potential costs of each action and selects the less expensive method to correct the situation. When the situation stabilizes and returns to normal, the system automatically resumes its hunt for a lower final dewatering cost. In summary, under normal operating conditions, the system will continuously search for ways to save money by varying polymer and decanter settings. When a process limit has been violated, the system will suspend the search for cost minimization and will resume it only when corrective action has been successfully taken. The end result is a smoother-running process, improved peace of mind for operators and lower overall cost. USING OCTOPUS AT MBC The automated optimization system used the San Diego MBC’s actual costs for four variables: polymer, cake disposal, solids recycling and power. The system uses this cost information as part of the software’s advanced decision algorithms. Octopus takes this information into consideration, along with any process constraints determined by MBC, to automatically select the appropriate polymer dose and decanter differential speed to deliver the lowest possible overall operational cost. The automated optimization system was also used to gather and store process data for both the optimized and manual centrifuge, allowing for a valid comparison. Results from the two decanters were then crosschecked and compared.

To further test the accuracy of the system readings, “grab samples” of sludge feed, centrate, and cake were taken and analyzed by both MBC lab staff and Alfa Laval technical staff throughout the test period.

RESULTS

The results of MBC lab samples were plotted against the process variables trended by the automated optimization system. On the first test, accuracy, averaged results from 57 grab samples taken during the test period were compared with the process variables historically trended by the autopilot system. The results showed high levels of accuracy. See Table 1. Table 1: Lab Test Analysis - System Accuracy

Averages over 57 Samples DC 5 DC7

Cake dryness (% DS) Octopus 29.58 28.17 MBC Lab 29.62 28.08 Diff Octopus vs. MBC -0.04 0.09

Centrate (mg/l) Octopus 462 502 MBC Lab 444 488 Diff Octopus vs. MBC 18 14

Feed (% DS) Octopus 2.03 AL Lab 1.99 MBC Lab n/a

Regarding costs, the automated operation resulted in a lower overall total cost than the manually run system. Because all operational costs are interrelated, a single parameter may be higher during automated optimization than with manual operation, while still offering overall reduced costs. This was the case during the test period at MBC. For example, during the test, centrate recovery costs of DC5 during automated optimization were slightly higher than those of the manually controlled decanter. However, this allowed the automated system to realize lower overall average costs for both cake dryness and polymer. In fact, during the test period, the facility saved an average of $3.51 per ton in operational costs when using automated vs. manual operation. See Table 2.

Table 2: Comparison of the combined costs per ton for DC5 and DC7 as run manually vs. optimized operation

Average of DC5 and DC7 Optimized $/ton DS Standby $/ton DS Difference Cake Dryness, % DS 29.2 $131.35 28.7 $133.79 ($2.44) Centrate, mg/l 463.1 $7.97 469.3 $8.07 ($0.11) Polymer, kg/ton 14.3 $20.67 14.9 $21.64 ($0.97) Total Cost $159.99 $163.50 ($3.51)

RESULTS As of the date of writing this paper, Alfa Laval Octopus is installed at 13 sites in Europe and three sites in North America (San Diego MBC, Hartford MDC, and Chicago-Calumet WRP). The local cost structure and any site-specific constraints will determine what process parameters the system will choose to optimize. In situations where disposal and/or transportation are major cost drivers, the high associated cost will make the system emphasize cake dryness. In cases where the disposal is a low cost, the system will drive the polymer cost to the lowest possible level. In all cases, the overriding principle is to reduce the total dewatering cost to the minimum possible under the site’s cost structure and other constraints. The system is customized to meet the specific requirements of each site so cost savings vary significantly from site to site. At San Diego MBC, the cost savings varied from a high of $5.71 per dry ton (DC 5) to a low of $1.04 (DC 7). The average overall savings of the two centrifuges was $3.51 per dry ton. CONCLUSIONS The results of the test satisfied both objectives of accuracy and savings at MBC. This will become an even more important tool given the cost increases expected due to projected future tougher environmental regulations and the increasing chemical, transportation and personnel costs. As a result of the test, MBC installed Octopus on four of their dewatering centrifuges and they continue to optimize their process with it today.

ACKNOWLEDGMENTS We would like to thank the personnel at San Diego MBC for their valuable time and expertise in carrying out the successful test of Alfa Laval Octopus and the subsequent permanent installation of the system on several of their centrifuges.