The primary objective of wastewater treatment is the removal of pollutants to meet given legal effluent standards. To further reduce operators costs additional recovery of resources and energy is desired by industrial and municipal wastewater treatment. Hence the objective in early stage of planning of treatment facilities lies in the identification and evaluation of promising configurations of treatment units. Obviously this early stage of planning may best be supported by software tools to be able to deal with a variety of different treatment configurations. In chemical process engineering various design tools are available that automatically identify feasible process configurations for the purpose to obtain desired products from given educts. In contrast, the adaptation of these design tools for the automatic generation of treatment unit configurations (process chains) to achieve preset effluent standards is hampered by the following three reasons. First, pollutants in wastewater are usually not defined as chemical substances but by compound parameters according to equal properties (e.g. all particulate matter). Consequently the variation of a single compound parameter leads to a change of related parameters (e.g. relation between Chemical Oxygen Demand and Total Suspended Solids). Furthermore, mathematical process models of treatment processes are tailored towards fractions of compound parameters. This hampers the generic representation of these process models which in turn is essential for automatic identification of treatment configurations. Second, treatment technologies for wastewater treatment rely on a variety of chemical, biological, and physical phenomena. Approaches to mathematically describe these phenomena cover a wide range of modeling techniques including stochastic, conceptual or deterministic approaches. Even more the consideration of temporal and spatial resolutions differ. This again hampers a generic representation of process models. Third, the automatic identification of treatment configurations may either be achieved by the use of design rules or by permutation of all possible combinations of units stored within a database of treatment units. The first approach depends on past experience translated into design rules. Hence, no innovative new treatment configurations can be identified. The second approach to identify all possible configurations collapses by extremely high numbers of treatment configurations that cannot be mastered. This is due to the phenomena of combinatorial explosion. It follows therefrom that an appropriate planning algorithm should function without the need of additional design rules and should be able to identify directly feasible configurations while discarding those impractical. This work presents a planning tool for the identification and evaluation of treatment configurations that tackles the before addressed problems. The planning tool comprises two major parts. An external declarative knowledge base and the actual planning tool that includes a goal oriented planning algorithm. The knowledge base describes parameters for wastewater characterization (i.e. material model) and a set of treatment units represented by process models (i.e. process model). The formalization of the knowledge base is achieved by the Web Ontology Language (OWL). The developed data model being the organization structure of the knowledge base describes relations between wastewater parameters and process models to enable for generic representation of process models. Through these parameters for wastewater characterization as well as treatment units can be altered or added to the knowledge base without the requirement to synchronize already included parameter representations or process models. Furthermore the knowledge base describes relations between parameters and properties of water constituents. This allows to track changes of all wastewater parameters which result from modeling of removal efficiency of applied treatment units. So far two generic treatment units have been represented within the knowledge base. These are separation and conversion units. These two raw types have been applied to represent different types of clarifiers and biological treatment units. The developed planning algorithm is based on a Means-Ends Analysis (MEA). This is a goal oriented search algorithm that posts goals from wastewater state and limit value restrictions to select those treatment units only that are likely to solve the treatment problem. Regarding this, all treatment units are qualified according to postconditions that describe the effect of each unit. In addition, units are also characterized by preconditions that state the application range of each unit. The developed planning algorithm furthermore allows for the identification of simple cycles to account for moving bed reactor systems (e.g. functional unit of aeration tank and clarifier). The evaluation of identified treatment configurations is achieved by total estimated cost of each configuration. The planning tool has been tested on five use cases. Some use cases contained multiple sources and sinks. This showed the possibility to identify water reuse capabilities as well as to identify solutions that go beyond end of pipe solutions. Beyond the originated area of application, the planning tool may be used for advanced interrogations. Thereby the knowledge base and planning algorithm may be further developed to address the objectives to identify configurations for any type of material and energy recovery.

The primary objective of wastewater treatment is the removal of pollutants to meet given legal effluent standards. To further reduce operators costs additional recovery of resources and energy is desired by industrial and municipal wastewater treatment. Hence the objective in early stage of planning of treatment facilities lies in the identification and evaluation of promising configurations of treatment units. Obviously this early stage of planning may best be supported by software tools to be able to deal with a variety of different treatment configurations. In chemical process engineering various design tools are available that automatically identify feasible process configurations for the purpose to obtain desired products from given educts. In contrast, the adaptation of these design tools for the automatic generation of treatment unit configurations (process chains) to achieve preset effluent standards is hampered by the following three reasons. First, pollutants in wastewater are usually not defined as chemical substances but by compound parameters according to equal properties (e.g. all particulate matter). Consequently the variation of a single compound parameter leads to a change of related parameters (e.g. relation between Chemical Oxygen Demand and Total Suspended Solids). Furthermore, mathematical process models of treatment processes are tailored towards fractions of compound parameters. This hampers the generic representation of these process models which in turn is essential for automatic identification of treatment configurations. Second, treatment technologies for wastewater treatment rely on a variety of chemical, biological, and physical phenomena. Approaches to mathematically describe these phenomena cover a wide range of modeling techniques including stochastic, conceptual or deterministic approaches. Even more the consideration of temporal and spatial resolutions differ. This again hampers a generic representation of process models. Third, the automatic identification of treatment configurations may either be achieved by the use of design rules or by permutation of all possible combinations of units stored within a database of treatment units. The first approach depends on past experience translated into design rules. Hence, no innovative new treatment configurations can be identified. The second approach to identify all possible configurations collapses by extremely high numbers of treatment configurations that cannot be mastered. This is due to the phenomena of combinatorial explosion. It follows therefrom that an appropriate planning algorithm should function without the need of additional design rules and should be able to identify directly feasible configurations while discarding those impractical. This work presents a planning tool for the identification and evaluation of treatment configurations that tackles the before addressed problems. The planning tool comprises two major parts. An external declarative knowledge base and the actual planning tool that includes a goal oriented planning algorithm. The knowledge base describes parameters for wastewater characterization (i.e. material model) and a set of treatment units represented by process models (i.e. process model). The formalization of the knowledge base is achieved by the Web Ontology Language (OWL). The developed data model being the organization structure of the knowledge base describes relations between wastewater parameters and process models to enable for generic representation of process models. Through these parameters for wastewater characterization as well as treatment units can be altered or added to the knowledge base without the requirement to synchronize already included parameter representations or process models. Furthermore the knowledge base describes relations between parameters and properties of water constituents. This allows to track changes of all wastewater parameters which result from modeling of removal efficiency of applied treatment units. So far two generic treatment units have been represented within the knowledge base. These are separation and conversion units. These two raw types have been applied to represent different types of clarifiers and biological treatment units. The developed planning algorithm is based on a Means-Ends Analysis (MEA). This is a goal oriented search algorithm that posts goals from wastewater state and limit value restrictions to select those treatment units only that are likely to solve the treatment problem. Regarding this, all treatment units are qualified according to postconditions that describe the effect of each unit. In addition, units are also characterized by preconditions that state the application range of each unit. The developed planning algorithm furthermore allows for the identification of simple cycles to account for moving bed reactor systems (e.g. functional unit of aeration tank and clarifier). The evaluation of identified treatment configurations is achieved by total estimated cost of each configuration. The planning tool has been tested on five use cases. Some use cases contained multiple sources and sinks. This showed the possibility to identify water reuse capabilities as well as to identify solutions that go beyond end of pipe solutions. Beyond the originated area of application, the planning tool may be used for advanced interrogations. Thereby the knowledge base and planning algorithm may be further developed to address the objectives to identify configurations for any type of material and energy recovery.