A full-scale operational digital twin for a water resource recovery facility-A case study of Eindhoven Water Resource Recovery Facility.

Digital transformation for the water sector has gained momentum in recent years, and many water resource recovery facilities modelers have already started transitioning from developing traditional models to digital twin (DT) applications. DTs simulate the operation of treatment plants in near real time and provide a powerful tool to the operators and process engineers for real-time scenario analysis and calamity mitigation, online process optimization, predictive maintenance, model-based control, and so forth. So far, only a few mature examples of full-scale DT implementations can be found in the literature, which only address some of the key requirements of a DT. This paper presents the development of a full-scale operational DT for the Eindhoven water resource recovery facility in The Netherlands, which includes a fully automated data-pipeline combined with a detailed mechanistic full-plant process model and a user interface co-created with the plant's operators. The automated data preprocessing pipeline provides continuous access to validated data, an influent generator provides dynamic predictions of influent composition data and allows forecasting 48 h into the future, and an advanced compartmental model of the aeration and anoxic bioreactors ensures high predictive power. The DT runs near real-time simulations every 2 h. Visualization and interaction with the DT is facilitated by the cloud-based TwinPlant technology, which was developed in close interaction with the plant's operators. A set of predefined handles are made available, allowing users to simulate hypothetical scenarios such as process and equipment failures and changes in controller settings. The combination of the advanced data pipeline and process model development used in the Eindhoven DT and the active involvement of the operators/process engineers/managers in the development process makes the twin a valuable asset for decision making with long-term reliability. PRACTITIONER POINTS: A full-scale digital twin (DT) has been developed for the Eindhoven WRRF. The Eindhoven DT includes an automated continuous data preprocessing and reconciliation pipeline. A full-plant mechanistic compartmental process model of the plant has been developed based on hydrodynamic studies. The interactive user interface of the Eindhoven DT allows operators to perform what-if scenarios on various operational settings and process inputs. Plant operators were actively involved in the DT development process to make a reliable and relevant tool with the expected added value.

Towards an online mitigation strategy for N2O emissions through principal components analysis and clustering techniques.

Emission of N2O represents an increasing concern in wastewater treatment, in particular for its large contribution to the plant's carbon footprint (CFP). In view of the potential introduction of more stringent regulations regarding wastewater treatment plants' CFP, there is a growing need for advanced monitoring with online implementation of mitigation strategies for N2O emissions. Mechanistic kinetic modelling in full-scale applications, are often represented by a very detailed representation of the biological mechanisms resulting in an elevated uncertainty on the many parameters used while limited by a poor representation of hydrodynamics. This is particularly true for current N2O kinetic models. In this paper, a possible full-scale implementation of a data mining approach linking plant-specific dynamics to N2O production is proposed. A data mining approach was tested on full-scale data along with different clustering techniques to identify process criticalities. The algorithm was designed to provide an applicable solution for full-scale plants' control logics aimed at online N2O emission mitigation. Results show the ability of the algorithm to isolate specific N2O emission pathways, and highlight possible solutions towards emission control.

The future of WRRF modelling - outlook and challenges.

The wastewater industry is currently facing dramatic changes, shifting away from energy-intensive wastewater treatment towards low-energy, sustainable technologies capable of achieving energy positive operation and resource recovery. The latter will shift the focus of the wastewater industry to how one could manage and extract resources from the wastewater, as opposed to the conventional paradigm of treatment. Debatable questions arise: can the more complex models be calibrated, or will additional unknowns be introduced? After almost 30 years using well-known International Water Association (IWA) models, should the community move to other components, processes, or model structures like 'black box' models, computational fluid dynamics techniques, etc.? Can new data sources - e.g. on-line sensor data, chemical and molecular analyses, new analytical techniques, off-gas analysis - keep up with the increasing process complexity? Are different methods for data management, data reconciliation, and fault detection mature enough for coping with such a large amount of information? Are the available calibration techniques able to cope with such complex models? This paper describes the thoughts and opinions collected during the closing session of the 6th IWA/WEF Water Resource Recovery Modelling Seminar 2018. It presents a concerted and collective effort by individuals from many different sectors of the wastewater industry to offer past and present insights, as well as an outlook into the future of wastewater modelling.

Tanks in series versus compartmental model configuration: considering hydrodynamics helps in parameter estimation for an N2O model.

The choice of the spatial submodel of a water resource recovery facility (WRRF) model should be one of the primary concerns in WRRF modelling. However, currently used mechanistic models are limited by an over-simplified representation of local conditions. This is illustrated by the general difficulties in calibrating the latest N2O models and the large variability in parameter values reported in the literature. The use of compartmental model (CM) developed on the basis of accurate hydrodynamic studies using computational fluid dynamics (CFD) can take into account local conditions and recirculation patterns in the activated sludge tanks that are important with respect to the modelling objective. The conventional tanks in series (TIS) configuration does not allow this. The aim of the present work is to compare the capabilities of two model layouts (CM and TIS) in defining a realistic domain of parameter values representing the same full-scale plant. A model performance evaluation method is proposed to identify the good operational domain of each parameter in the two layouts. Already when evaluating for steady state, the CM was found to provide better defined parameter ranges than TIS. Dynamic simulations further confirmed the CM's capability to work in a more realistic parameter domain, avoiding unnecessary calibration to compensate for flaws in the spatial submodel.

Multi-point monitoring of nitrous oxide emissions in three full-scale conventional activated sludge tanks in Europe.

The large global warming potential of nitrous oxide (N2O) is currently of general concern for the water industry, especially in view of a new regulatory framework concerning the carbon footprint of water resource recovery facilities (WRRFs). N2O can be generated through different biological pathways and from different treatment steps of a WRRF. The use of generic emission factors (EF) for quantifying the emissions of WRRFs is discouraged. This is due to the number of different factors that can affect how much, when and where N2O is emitted from WRRFs. The spatial and temporal variability of three WRRFs in Europe using comparable technologies is presented. An economically feasible and user-friendly method for accounting for the contribution of anoxic zones via direct gas emission measurements was proven. The investigation provided new insights into the contribution from the anoxic zones versus the aerobic zones of biological WRRF tanks and proved the unsuitability of the use of a single EF for the three WRRFs. Dedicated campaigns for N2O emissions assessment are to be advised. However, similarities in the EF magnitude can be found considering treatment strategy and influent water composition.

Assessing Urban Wastewater System Upgrades Using Integrated Modeling, Life Cycle Analysis, and Shadow Pricing.

This study assesses the environmental impacts of four measures proposed for upgrading of the urban wastewater system of Eindhoven and the Dommel River in The Netherlands, against the base case, "do-nothing" option. The measures aim to reduce the overall environmental impact of the Eindhoven urban wastewater system (UWS) by targeting river dissolved oxygen depletion and ammonia peaks, reducing combined sewer overflows, and enhancing nutrient removal. The measures are evaluated using a life cycle analysis with the boundaries including the receiving river section by means of an integrated model of the UWS. An uncertainty analysis of the estimated impacts has been performed to support the outcomes. The study also uses the economic concept of shadow prices to assign relative weights of socio-economic importance to the estimated life cycle impacts. This novel integration of tools complements the assessments of this UWS with the inclusion of long-term global environmental impacts and the investigation of trade-offs between different environmental impacts through a single monetary unit. The results support the selection of deeper clarifiers as the most environmentally beneficial measure for upgrade.

Cost-effective solutions for water quality improvement in the Dommel River supported by sewer-WWTP-river integrated modelling.

This project aims at finding cost-efficient sets of measures to meet the Water Framework Directive (WFD) derived goals for the Dommel River (The Netherlands). Within the project, both acute and long-term impacts of the urban wastewater system on the chemical and ecological quality of the river are studied with a monitoring campaign in the urban wastewater system (wastewater treatment plant and sewers) and in the receiving surface water system. An integrated model, which proved to be a powerful tool to analyse the interactions within the integrated urban wastewater system, was first used to evaluate measures in the urban wastewater system using the existing infrastructure and new real-time control strategies. As the latter resulted to be beneficial but not sufficient, this paper investigated the use of additional infrastructural measures to improve the system cost-effectively and have it meet the Directive's goals. Finally, an uncertainty analysis was conducted to investigate the impact of uncertainty in the main model assumptions and model parameters on the performance robustness of the selected set of measures. Apart from some extreme worst-case scenarios, the proposed set of measures turned out to be sufficiently robust. Due to the substantial savings obtained with the results of this project, the pay-back time of the whole monitoring and modelling work proved to be less than 5 months. This illustrates the power of mathematical modelling for decision support in the context of complex urban water systems.

Modelling and monitoring of integrated urban wastewater systems: review on status and perspectives.

While the general principles and modelling approaches for integrated management/modelling of urban water systems already present a decade ago still hold, in recent years aspects like model interfacing and wastewater treatment plant (WWTP) influent generation as complements to sewer modelling have been investigated and several new or improved systems analysis methods have become available. New/improved software tools coupled with the current high computational capacity have enabled the application of integrated modelling to several practical cases, and advancements in monitoring water quantity and quality have been substantial and now allow the collecting of data in sufficient quality and quantity to permit using integrated models for real-time applications too. Further developments are warranted in the field of data quality assurance and efficient maintenance.

On data requirements for calibration of integrated models for urban water systems.

Modeling of integrated urban water systems (IUWS) has seen a rapid development in recent years. Models and software are available that describe the process dynamics in sewers, wastewater treatment plants (WWTPs), receiving water systems as well as at the interfaces between the submodels. Successful applications of integrated modeling are, however, relatively scarce. One of the reasons for this is the lack of high-quality monitoring data with the required spatial and temporal resolution and accuracy to calibrate and validate the integrated models, even though the state of the art of monitoring itself is no longer the limiting factor. This paper discusses the efforts to be able to meet the data requirements associated with integrated modeling and describes the methods applied to validate the monitoring data and to use submodels as software sensor to provide the necessary input for other submodels. The main conclusion of the paper is that state of the art monitoring is in principle sufficient to provide the data necessary to calibrate integrated models, but practical limitations resulting in incomplete data-sets hamper widespread application. In order to overcome these difficulties, redundancy of future monitoring networks should be increased and, at the same time, data handling (including data validation, mining and assimilation) should receive much more attention.

The incorporation of variability and uncertainty evaluations in WWTP design by means of stochastic dynamic modeling: the case of the Eindhoven WWTP upgrade.

This paper illustrates how a dynamic model can be used to evaluate a plant upgrade on the basis of post-upgrade performance data. The case study is that of the Eindhoven wastewater treatment plant upgrade completed in 2006. As a first step, the design process based on a static model was thoroughly analyzed and the choices regarding variability and uncertainty (i.e. safety factors) were made explicit. This involved the interpretation of the design guidelines and other assumptions made by the engineers. As a second step, a (calibrated) dynamic model of the plant was set up, able to reproduce the anticipated variability (duration and frequency). The third step was to define probability density functions for the parameters assumed to be uncertain, and propagate that uncertainty with the dynamic model by means of Monte Carlo simulations. The last step was the statistical evaluation and interpretation of the simulation results. This work should be regarded as a 'learning exercise' increasing the understanding of how and to what extent variability and uncertainty are currently incorporated in design guidelines used in practice and how model-based post-project appraisals could be performed.

Impact of influent data frequency and model structure on the quality of WWTP model calibration and uncertainty.

Application of activated sludge models (ASMs) to full-scale wastewater treatment plants (WWTPs) is still hampered by the problem of model calibration of these over-parameterised models. This either requires expert knowledge or global methods that explore a large parameter space. However, a better balance in structure between the submodels (ASM, hydraulic, aeration, etc.) and improved quality of influent data result in much smaller calibration efforts. In this contribution, a methodology is proposed that links data frequency and model structure to calibration quality and output uncertainty. It is composed of defining the model structure, the input data, an automated calibration, confidence interval computation and uncertainty propagation to the model output. Apart from the last step, the methodology is applied to an existing WWTP using three models differing only in the aeration submodel. A sensitivity analysis was performed on all models, allowing the ranking of the most important parameters to select in the subsequent calibration step. The aeration submodel proved very important to get good NH(4) predictions. Finally, the impact of data frequency was explored. Lowering the frequency resulted in larger deviations of parameter estimates from their default values and larger confidence intervals. Autocorrelation due to high frequency calibration data has an opposite effect on the confidence intervals. The proposed methodology opens doors to facilitate and improve calibration efforts and to design measurement campaigns.

Developing a framework for continuous use of models in daily management and operation of WWTPs: a life cycle approach.

We developed and evaluated a framework for the continuous use of dynamic models in daily management and operation of WWTPs. The overall aim is to generate knowledge and build in-house capacity for the reliable use of dynamic models in practice (within a regional water authority in The Netherlands). To this end, we have adopted a life cycle approach, where the plant model follows the different stages that make up the typical lifespan of a plant. Since this approach creates a framework in which models are continuously reused, it is more efficient in terms of resources and investment than the traditional approach where one always makes a new model for the plant whenever it is needed. The methodology was evaluated successfully at a 50,000 PE domestic EBPR plant (Haaren, The Netherlands). It is shown that the continuous use and update of models in a cyclic manner creates a learning cycle, which results in experience and knowledge generation about the plant's modelling that accumulates and translates into improvements into the modelling quality and efficiency. Moreover, a model is now always on-the-shelf for process optimization.

An efficient approach to automate the manual trial and error calibration of activated sludge models.

An efficient approach is introduced to help automate the rather tedious manual trial and error way of model calibration currently used in activated sludge modeling practice. To this end, we have evaluated a Monte Carlo based calibration approach consisting of four steps: (i) parameter subset selection, (ii) defining parameter space, (iii) parameter sampling for Monte Carlo simulations and (iv) selecting the best Monte Carlo simulation thereby providing the calibrated parameter values. The approach was evaluated on a formerly calibrated full-scale ASM2d model for a domestic plant (located in The Netherlands), using in total 3 months of dynamic oxygen, ammonia and nitrate sensor data. The Monte Carlo calibrated model was validated successfully using ammonia, oxygen and nitrate data collected at high measurement frequency. Statistical analysis of the residuals using mean absolute error (MAE), root mean square error (RMSE) and Janus coefficient showed that the calibrated model was able to provide statistically accurate and valid predictions for ammonium, oxygen and nitrate. This shows that this pragmatic approach can perform the task of model calibration and therefore be used in practice to save the valuable time of modelers spent on this step of activated sludge modeling. The high computational demand is a downside of this approach but this can be overcome by using distributed computing. Overall we expect that the use of such systems analysis tools in the application of activated sludge models will improve the quality of model predictions and their use in decision making.