Use of a Semi-Batch Process Centrifugation Process to Capture and Concentrate Finfish Effluent, a Platform Technology Suitable for Many Aquaculture Production Systems

AIMAP 2009-P34

Table of contents

• Background
• Objectives
• Results of the Feed Dewatering Tiral
• Discussion and Next Steps
• Addendum
• Separation efficiency and cost effectiveness
• Desalination Efficiency

Background

Agrimarine Industries Inc. develops and operates solid-walled floating closed-containment finfish aquaculture systems for salmon and other species. Agrimarine closed- containment enclosures incorporate a dual-drain solid waste recovery feature which removes over 80% of feces and uneaten feed from the culture water prior to discharge into the marine environment. Separation is driven by a constant 20 liter per second pumped flow (the primary stream), resulting in a maximum of 1% solids. The primary stream is too dilute and massive for direct  land application, and  must  be further dewatered and rinsed prior to composting. The current project will develop methods and equipment for dewatering and partially desalinating the primary stream by means of semi- batch filter centrifugation, resulting in a solid cake suitable for direct land application.

After consideration of several alternatives, it was decided that a two-stage separation system  would result in the greatest energy efficiency. The primary stream will be siphoned from the dual-drain into a floating vertical column gravity separator, moored adjacent to the rearing enclosure. Simple gravity separation will produce sludge with expected solids content of 2-5%, and clear overflow water. The sludge (the secondary stream) will then be intermittently pumped from the gravity separator to the final dewatering stage. Use of a floating primary separator will reduce the pumping volume requirement of the secondary stream, and reduce the final centrate volume.

A vertical axis filter centrifuge was chosen for final dewatering and desalination. Filter centrifugation uses both density differences and water permeability of the solids to effect separation, and can potentially be achieved at much lower centrifugal field than solid bowl centrifugation. Filter centrifuges also allow the cake to be rinsed while spinning, without re-suspension of solids. Rinsing with fresh water will allow partial de-salination of the cake, reducing chloride contamination of soil following land application. As new filter centrifuges are expensive, three used machines were purchased for this project. They were in ‘as-is’ condition, and were expected to require some refurbishment prior to successful use.

At the time of this writing, the installation of the floating closed-containment aquaculture facility on which this project depends was behind schedule, and salmon smolts had only recently been stocked. A preliminary trial was therefore conducted to assess the efficiency of the centrifuge in separating solids from a slurry of salmon feed in seawater. The slurry simulates the uneaten feed component of the primary stream. Separation time and centrifuge cake characteristics were determined in order to provide information needed for selection of a permanent filter screen, to be installed during the solids dewatering trial.

Objectives

1. Verify correct operation of the first filter centrifuge
2. Determine dewatering time for feed-seawater mixtures

Salmon feed pellets (Taplow Chinook Grower, 8mm) were weighed into measured, 100L volumes of ambient  seawater, (Middle Bay, BC 29 ppt salinity) in polyethylene containers. Feed was soaked overnight, then slurried with an electric drill and paint mixer attachment (250 RPM, 5 minutes). A filter centrifuge (Comi-Condor ALPHA/SB-800), equipped with internal rinse spray-bar was fitted with a nylon fabric filter bag. Slurry mixtures were pumped into the centrifuge with a portable pump while the centrifuge was operated at low speed. Loading time and centrifuge speed were recorded. After loading was complete, centrate flow was monitored while the centrifuge was maintained at constant speed, and time until cessation of centrate flow was recorded. Fresh tap-water was then admitted through the spray-bar. Rinse centrate was then sampled for salinity measurement. Due to a procedural error in salinity determination, desalination efficiency will not be included in this report. Centrifuge speed was then increased to achieve final cake dewatering, and time until cessation of centrate flow was recorded.

Final cake thickness and weight were recorded and compared with initial feed sample weight.

Results of the Feed Dewatering Tiral

Due to difficulties with tearing of the fabric filter bag, loading speed was lower than could have been achieved, and full speed was not used for the final rinse. Consequently, total cycle time was longer than expected, as high as 39 minutes. Cake thickness was higher at the bottom of the drum than the top, although the product loading spout discharges at mid-height on the drum, indicating that discharge from the spout was not distributed evenly. Centrifugal field at the recorded speed of 445 RPM was 88G, so gravitational settling was an unlikely cause of the uneven cake thickness. The dewatered and rinsed cake was compact and friable, and could easily be handled by shovel.

Discussion and Next Steps

Highest anticipated production volume shared by two machines (alternating loading and rinse/dewater portions of the cycle) will require cycle times of 15 minutes. Therefore significant reduction in loading, rinse, and final dewatering time will be required. Water permeability of fecal solids is expected to be lower than that of feed slurry, so higher operating speeds will be  required. The fabric bag filter medium is not suitable for automated operation of the centrifuge, as bottom discharge operation will be required. A wedge-wire or slotted plastic filter screen and hydraulic rake discharge apparatus will be installed. A more rugged screen will withstand greater loading flow and allow full speed operation of 1800 RPM, corresponding to 90% greater centrifugal field. Determination of solids separation efficiency will require analysis of primary stream overflow and centrate total solids concentration, as well as a dry weight determination of the centrifuge cake. Determination of de-salination effectiveness can be accomplished by conductivity analysis of centrate at various points of the rinse cycle, and sodium determination of the dewatered cake.

At this writing, the rearing enclosure has been stocked with 56,000 Chinook salmon smolts, and feeding rate is about 30 kg/day. The centrifuge has been connected directly to the rearing tank drain, allowing intermittent operation. The first filter centrifuge (used in this trial) will remain in operation until machines 2 and 3 are rebuilt and installed as an automated system.

Future update reports on this project will focus on desalination, total separation efficiency, and dry weight determination of the centrifuge cake.

Addendum:

Separation efficiency and cost effectiveness

Based on the observed centrifuge cycle times and average motor current it is possible to develop a preliminary estimate of operating cost of the centrifuge waste separation system. While the test material was not expected to behave exactly as actual waste (mixed feed and  fecal waste), the initial dewatering  time of 300 seconds is consistent with the dewatering time requirement of a commercial-scale centrifuge. At a total feeding rate of 10 tonnes per day, and dry weight fecal solids production of 20% of feed fed, 6700 kg of 70% moisture centrifuge cake will be produced. Assuming this can be fed to the centrifuge at a steady rate of 279 kg/hr, 3 cycles each dewatering 90 kg of solids cake would be required per hour. As this trial achieved loading, rinse, and final dewatering within 25 minutes in all trials, a 20 minute cycle in future seems achievable.

The rotational energy delivered to the centrifuge feed is directly proportional to rotation angular velocity, so the typical dewatering speed of 750 RPM was expected to require about 50% of the rated power consumption of the drive motor, or about 6.5 kW. This was roughly the observed actual consumption. In contrast to the filter centrifuge used in this trial, a solid bowl centrifuge would operate at full speed throughout the separation process, and the water contained in the centrifuge feed (about 98%) would be accelerated to the maximum operating speed. The filter centrifuge therefore provides a potential energy saving of about 50% relative to a solid-bowl machine, with the added benefit of fresh-water rinse.

Waste separation efficiency at the centrifuge was determined by total suspended solids analysis of the centrate. The centrifuge feed is expected to average 1% solids, while the centrate from this trial contained 10 mg/L of TSS. The separation efficiency of the entrifuge was therefore 99.9%. The project goal of 80% solids separation efficiency refers principally to that of the in-tank waste trap. Efficient operation of the trap will require continuous operation of the primary separator and higher biomass loading. At the time of this writing, the primary separator was under construction and the rearing tank had only been stocked with smolts for a few weeks, so fecal production rate was still too low to provide an accurate estimate of waste trap performance.

Overall operating cost of the waste system can be estimated based on this trial. A single centrifuge, operating at 6.5 kW (50% rated power) and one primary sludge pump operating at 3 kW would be required for a commercial scale farm. The electricity operating cost of the system would be about 60 cents per hour, or $400 per peak month. By comparison, feed cost during that month would total about $540,000.

Cost recovery from sales or other profitable use of the dewatered solids is possible, and will be investigated when sufficient material is available. Possible uses include household bagged garden compost (sold through retail), agricultural land development from freshly cleared forest, or chemical processing to extract concentrated phosphate.

Desalination Efficiency:

Salinity of the centrifuged cake was determined by re-suspending 500g of centrifuge cake with 1L of fresh water, so that the salinity of the pore water within the cake is expected to have been 3 times greater than the salinity listed. Rinsing therefore appeared to achieve a 5 to 10-fold reduction in salt concentration. This level of salinity would be acceptable for direct land application on Vancouver Island, as further dilution on mixing with soil would result in salt levels well within acceptable levels. Further investigation using fecal solids will be required to further reduce salinity while minimizing fresh water use in the process.