Search

Abstract: Fish production in marine netcages is expanding rapidly in many parts of the world. Nevertheless, production planning and layout design tools still rely on rule of thumb, spreadsheets, local experience and practices handed down through the generations. An integrated model (queuing, optimization and simulation were linked together) was developed. The solution finds the values of decision variables to maximize yearly production. The concept was that a single netcage can be seen as a “server” in which neither a “queue” (over-holding of fish), nor an idle netcage is allowed. A marine fish farm can then be seen as a queuing network, and a queuing network-based management model was developed. Growth data of 40 batches of fish, each batch comprising on average 180,000 fish (std 50,000), were recorded over an experimental period of 4 years. The model inputs were (1) the empirical fish growth rates (2) the given space for netcages, (3) preferred netcage holes, netcage depth and netcage diameter, (4) fingerling supply limitations, (5) market timing, and (6) Preferred market-size fish at the farm gate. The model outputs were: (7) optimal fingerling arrival frequency, (8) optimal number of fingerlings in a batch, (9) number of days in each culture netcage, (10) grading and sorting criteria along the production line, and (11) optimal facility allocation (number of netcages for each growing phase). Model validity was statistically tested and was not rejected within the 95% confidence level. The model application results with 4 netcages in the 1st growing phase, 8 netcages in the 2nd growing phase and 16 netcages in the 3rd growing phase (so called “4,8,16 layout”) gave the following optimal operating parameters: arrival of a batch every 30 days; 122 days in each successive growth phase. The optimal values satisfied the biomass density criterion of less than 25kgm−3 and the netcage utilization criterion of never below 99%. Expected production was 2403tonyear−1 (vs. the current 686tonyear−1). The enterprise owners decided to adopt the model results and the system is now being built according to the 4,8,16 design. [Copyright &y& Elsevier]

Abstract: The design and management of a recirculating aquaculture system (RAS) is crucial for the farm''s economic survival. In a previous paper (Part I of this study), a model was developed. The current paper extends the principles developed in Part I by (1) addressing a larger-scale RAS, (2) addressing the layout positioning problem, (3) integrating a robust 6σ design into the optimization problem. A queuing model and a solvable nonlinear constrained optimization problem including the 6σ robust design were developed and validated. The design criteria were: (1) turnover ≥1000ton/year, (2) 7 days quarantine, i.e., at least 7 days between arrivals of two successive fish batches, (3) fish biomass density ≤55kg/m3, (4) three growth phases, (5) neither fish-sorting nor batch-splitting events allowed, and (6) a robust design to accommodate two species—seabream and seabass grouper, with different growth rates. Decision variables were: (1) number of culture tanks, (2) fingerling arrival frequency, (3) number of fingerlings per batch, (4) number of days in a growth phase, (5) timing of grading and sorting criteria on the production lines, (5) standing biomass in the entire system, which is the actual biomass load on the biofilters, (6) feed amount per day. The optimal layout was: 13 culture tanks in each of the three growth phases (39 tanks total). Optimal parameters included: arrival frequency—a single fish batch into the system every 7 days, 91 days in each phase; growth up to 77, 233, and 468g in successive growth phases. Optimal values satisfied the criteria of biomass density below 50kg/m3 and culture tank utilization above 99%. Expected production was 1000ton/year. The proposed layout can accommodate different fish species—here, seabream and grouper—under the same culture volume, density, and schedule, but with different growth rates. Increasing the desired biomass density from 50 to 60kg/m3 advances expected production to 1335ton/year. The numerical values reflect local aquatic conditions, but the proposed methodology can be applied anywhere. [Copyright &y& Elsevier]

Abstract: This study describes the design and management of an effective recirculating aquaculture system (RAS). The RAS design involves many aspects, both physical and biological: (1) a desired turnover, (2) fingerling arrival frequency, (3) number of fingerlings per batch, (4) number of days in a growth phase, (4) timing of grading and sorting, based on (5) fish growth rate, and (6) number of culture tanks. The design criteria were: (1) turnover of 250ton/year, (2) fingerling arrival frequency of 12batches/year, (3) biomass density ≤60kg/m3, (4) two fish batch-sorting and batch-splitting events, and (5) a robust design to accommodate two species—slower- and faster-growing species. The culture tank was regarded as a queuing system in which neither a “queue” (overholding of fish) nor an idle culture tank is allowed, enabling modeling of the fish farm as a queuing network. A queuing model, stochastic simulation, optimization, and six sigma robust design were developed, validated, and implemented. The optimal layout was found to comprise three growth phases, with 1, 8, and 24 culture tanks, respectively. Optimal parameters included: arrival frequency—a single fish batch into the system every 30 days; then 30, 120 and 180 days in the 1st, 2nd and 3rd phases, to 42, 200, and 440g, respectively. The optimal values satisfied the criteria of biomass density below 60kg/m3 and culture tank utilization above 93%. Expected production was 250–276ton/year. The proposed layout can accommodate different fish species with different growth rates under the same proposed layout, culture volume, density, and schedule. The numerical values reflect local aquatic conditions, but the proposed methodology can be applied elsewhere. [Copyright &y& Elsevier]