Abstract Recirculating aquaculture systems (RAS) are often designed using simplified steady-state mass balances, which fail to account for the complex dynamics that biological water treatment systems exhibit. Because of the very slow dynamics, experimental development is also difficult. We present a new, fast and robust Modelica implementation of a material balance-based dynamic simulator for fish growth, waste production and water treatment in recirculating aquaculture systems. This simulator is used together with an optimization routine based on a genetic algorithm to evaluate the performance of three different water treatment topologies, each for two fish species (Rainbow trout and Atlantic salmon) and each in both a semi-closed (no denitrification) and a fully recirculating version (with denitrification). Each case is furthermore evaluated at both saturated and supersaturated oxygen levels in the fish tank influent. The 24 cases are compared in terms of volume required to maintain an acceptable TAN concentration in the fish tank. The results indicate that the smallest volume is obtainable by introducing several bypass flows in the treatment system of a semi-closed RAS and that the gains can be significant. We also show that recycling already treated water back upstream in the treatment process degrades performance and that if one wishes to have a fully recirculating system with minimal water exchange, then the flows of oxygen, carbon and nitrogen must be carefully considered. For several of the cases, no optimum with denitrification could be found. We thus demonstrate that the best configuration and operation strategy for water treatment varies with the conditions imposed by the fish culture, illustrating the complexity of RAS plants and the importance of simulations, but also that computer-driven optimal design has the potential to increase the treatment efficiency of biofilters which could lead to cheaper plants with better water quality.

中文翻译：

---

使用模拟和优化比较循环水产养殖系统的拓扑结构

摘要 循环水产养殖系统 (RAS) 通常使用简化的稳态质量平衡进行设计，这无法解释生物水处理系统表现出的复杂动态。由于动力学非常缓慢，实验开发也很困难。我们展示了一种新的、快速且强大的 Modelica 实现，它基于材料平衡的动态模拟器用于循环水产养殖系统中的鱼类生长、废物产生和水处理。该模拟器与基于遗传算法的优化程序一起使用，以评估三种不同水处理拓扑的性能，每种拓扑用于两种鱼类（虹鳟鱼和大西洋鲑鱼），每种都用于半封闭（无反硝化）和一个完全再循环的版本（带反硝化）。此外，每种情况都在鱼缸进水中的饱和和过饱和氧水平下进行评估。根据维持鱼缸中可接受的 TAN 浓度所需的体积来比较这 24 个案例。结果表明，通过在半封闭 RAS 的处理系统中引入几个旁路流可以获得最小的体积，并且收益可能很大。我们还表明，在处理过程中将已经处理过的水循环回上游会降低性能，如果希望拥有一个具有最少水交换的完全再循环系统，则必须仔细考虑氧气、碳和氮气的流动。对于一些情况，没有找到最佳的反硝化。