Search

Better controlling and optimising the plant's processes has become a priority for WWTP (Wastewater Treatment Plant) managers. The main objective of this project is to develop a simplified mathematical tool able to reproduce and anticipate the behaviour of the Tougas WWTP (Nantes, France). This tool is aimed to be used directly by the managers of the site. The mathematical WWTP model was created using the software WEST. This paper describes the studied site and the modelling results obtained during the stage of the model calibration and validation. The good simulation results have allowed to show that despite a first very simple description of the WWTP, the model was able to correctly predict the nitrogen composition (ammonia and nitrate) of the effluent and the daily sludge extraction. Then, a second more detailed configuration of the WWTP was implemented. It has allowed to independently study the behaviour of each of four biological trains. Once this first stage will be completely achieved, the remainder of the study will focus on the operational use of a simplified simulator with the purpose of optimising the Tougas WWTP operation.

Membrane bioreactors (MBRs) are attracting global interest but the mathematical modeling of the biological performance of MBRs remains very limited. This study focuses on the modelling of a side-stream MBR system using Activated Sludge Model No. 1 (ASM1), and comparing the results with the modelling of traditional activated sludge processes. ASM1 parameters relevant for the long-term biological behaviour in MBR systems were calibrated (i.e. Y(H) = 0.72gCOD/gCOD, Y(A) = 0.25gCOD/gN, b(H) = 0.25 d(-1), b(A) = 0.080 d(-1) and f(P) = 0.06), and generally agreed with the parameters in traditional activated sludge processes, with the exception that a higher autotrophic biomass decay rate was observed in the MBR. A sensitivity analysis for steady state operation and DO dynamics suggested that the biological performance of the MBR system (the sludge concentration, effluent quality and the DO dynamics) are very sensitive to the parameters(i.e. Y(H), Y(A), b(H), b(A), micro(maxH) and micro(maxA)), and influent wastewater components(X(I), S(S), X(S), S(NH)).

The COST/IWA benchmark simulation model has been available for seven years. Its primary purpose has been to create a platform for control strategy benchmarking of biological wastewater treatment processes. The fact that the benchmark has resulted in more than 100 publications, not only in Europe but also worldwide, demonstrates the interest for such a tool in the research community. In this paper, an extension of the benchmark simulation model no. 1 (BSM1) is proposed. It aims at facilitating evaluation of two closely related operational tasks: long-term control strategy performance and process monitoring performance. The motivation for the extension is that these two tasks typically act on longer time scales. The extension proposed here consists of 1) prolonging the evaluation period to one year (including influent files), 2) specifying time varying process parameters and 3) including sensor and actuator failures. The prolonged evaluation period is necessary to obtain a relevant and realistic assessment of the effects of such disturbances. Also, a prolonged evaluation period allows for a number of long-term control actions/handles that cannot be evaluated in a realistic fashion in the one week BSM1 evaluation period. In the paper, models for influent file design, parameter changes and sensor failures, initialization procedure and evaluation criteria are discussed. Important remaining topics, for which consensus is required, are identified. The potential of a long-term benchmark is illustrated with an example of process monitoring algorithm benchmarking.

An improved methodology based on a nitrate biosensor is developed and applied successfully for in-depth monitoring and study of anoxic activated sludge activities. The major advantages of the methodology are its simplicity, reliability and high data quality. The resulting data allowed for the first time to monitor anoxic respiration rate of activated sludge (nitrate uptake rate (NUR)) at a high time resolution making it clearly comparable with high frequency oxygen uptake rate (OUR) measurements obtained under aerobic conditions. Further, the anoxic respiration data resulting from a pulse addition of carbon source to endogenously respiring anoxic activated sludge shows a clear start-up phenomenon and storage tail that is usually also observed in high-frequency OUR measurements. Finally, the improved methodology can be expected to serve as an anoxic respirometer for activated sludge treatment plants where denitrification process occurs in single-step. Further, it can be used for a variety of purposes e.g. for toxicity and activity monitoring, process control and parameter estimation of the activated sludge process, similar to the aerobic respirometers.

A systematic approach to determine the optimal operation strategy for nitrogen (N) and phosphorus (P) removal of sequencing batch reactors (SBRs) has been developed and applied successfully to a lab-scale SBR. The methodology developed is based on using a grid of possible scenarios to simulate the effect of the key degrees of freedom in the SBR system. The grid of scenarios is simulated using a calibrated ASM2dN model developed and calibrated in a previous study. Effluent quality in combination with a robustness index for each of the scenarios is used to select the best scenario. With the best scenario, it is possible to improve/increase the current performance of the SBR system by around 54% and 74% for N and P removal respectively.

A new approach to nonlinear modeling and adaptive monitoring using fuzzy principal component regression (FPCR) is proposed and then applied to a real wastewater treatment plant (WWTP) data set. First, principal component analysis (PCA) is used to reduce the dimensionality of data and to remove collinearity. Second, the adaptive credibilistic fuzzy-c-means method is used to appropriately monitor diverse operating conditions based on the PCA score values. Then a new adaptive discrimination monitoring method is proposed to distinguish between a large process change and a simple fault. Third, a FPCR method is proposed, where the Takagi-Sugeno-Kang (TSK) fuzzy model is employed to model the relation between the PCA score values and the target output to avoid the over-fitting problem with original variables. Here, the rule bases, the centers and the widths of TSK fuzzy model are found by heuristic methods. The proposed FPCR method is applied to predict the output variable, the reduction of chemical oxygen demand in the full-scale WWTP. The result shows that it has the ability to model the nonlinear process and multiple operating conditions and is able to identify various operating regions and discriminate between a sustained fault and a simple fault (or abnormalities) occurring within the process data.

This note presents and compares two models to predict the equilibrium temperature in aerated basins. They differ by their degree of complexity and therefore by the input data they require. Both models were able to estimate the temperature of an industrial aerated lagoon, the more complex model giving, in addition, a complete breakdown of the heat exchanges.

In probabilistic environmental risk assessment, the likelihood and the extent of adverse effects occurring in ecological systems because of exposure(s) to substances are estimated. It is based on the comparison of an exposure/environmental concentration distribution, with a species sensitivity distribution derived from toxicity data. The calculation of a probabilistic risk can be performed in many ways (e.g., area under the curve in joint probability curves). However, several (hypothetical) examples and some theoretical considerations illustrate that the current risk characterisation methods have an integrative character and they focus on the statistical comparison of two distributions without properly considering the environmental interpretation of these underlying distributions. Several scenarios with varying exposure/environmental concentration distribution and species sensitivity distribution standard deviations are discussed.

The first step in the estimation of parameters of models applied for data interpretation should always be an investigation of the identifiability of the model parameters. In this study the structural identifiability of the model parameters of Monod-based activated sludge models (ASM) was studied. In an illustrative example it was assumed that respirometric (dissolved oxygen or oxygen uptake rates) and titrimetric (cumulative proton production) measurements were available for the characterisation of nitrification. Two model structures, including the presence and absence of significant growth for description of long- and short-term experiments, respectively, were considered. The structural identifiability was studied via the series expansion methods. It was proven that the autotrophic yield becomes uniquely identifiable when combined respirometric and titrimetric data are assumed for the characterisation of nitrification. The most remarkable result of the study was, however, that the identifiability results could be generalised by applying a set of ASM1 matrix based generalisation rules. It appeared that the identifiable parameter combinations could be predicted directly based on the knowledge of the process model under study (in ASM1-like matrix representation), the measured variables and the biodegradable substrate considered. This generalisation reduces the time-consuming task of deriving the structurally identifiable model parameters significantly and helps the user to obtain these directly without the necessity to go too deeply into the mathematical background of structural identifiability.

Multiway principal component analysis (MPCA) for the analysis and monitoring of batch processes has recently been proposed. Although MPCA has found wide applications in batch process monitoring, it assumes that future batches behave in the same way as those used for model identification. In this study, a new monitoring algorithm, adaptive multiblock MPCA, is developed. The method overcomes the problem of changing process conditions by updating the covariance structure recursively. A historical set of operational data of a multiphase batch process was divided into local blocks in such a way that the variables from one phase of a batch run could be blocked in the corresponding blocks. This approach has significant benefits because the latent variable structure can change for each phase during the batch operation. The adaptive multiblock model also allows for easier fault detection and isolation by looking at the relationship between blocks and at smaller meaningful block models, and it therefore helps in the diagnosis of the disturbance. The proposed adaptive multiblock monitoring method is successfully applied to a sequencing batch reactor for biological wastewater treatment., (Copyright 2003 Wiley Periodicals, Inc. Biotechnol Bioeng 82: 489-497, 2003.)

Over a period of 227 days properties of activated sludge grown in an sequencing batch reactor (SBR) operated under stable conditions were analyzed. Settling properties (sludge volume index (SVI)) of the activated sludge were compared with on-line measurements of floc size and size distribution obtained by using a laser light scattering technique (Malvern Mastersizer/S, Malvern, UK), and with measurements of microbial community dynamics analyzed by denaturing gradient gel electrophoresis (DGGE) patterns of 16S rRNA genes. In addition, microscopical observations were used to confirm the results. Three distinct stages in the SBR evolution were observed. In the first stage the structural floc properties showed predominant presence of floc-forming bacteria in the activated sludge. A good correlation between floc size, properties and microbial community evolution was observed. The second stage showed a good balance between floc-forming and filamentous bacteria, with good settling properties and a highly dynamic community in the SBR. In the third stage, an increase in the filamentous bacteria, which became predominant in the system was observed. Again, a good correlation between settling properties and floc size distribution was obtained and a new dominant species was observed in the DGGE patterns, which can be assumed to be a filamentous organism.

A new one-dimensional clarifier model was developed, including components of existing models, and extended with a height-dependent cross-sectional area and two flowrate-dependent dispersion coefficients. This model is evaluated using data from a detailed one-month measuring campaign on a full-scale wastewater treatment plant. The data included hourly sludge concentration profile, sludge heights at 10 minute intervals, sludge concentrations in inlet, effluent and recycle flows and regular settling properties characterised by batch setting tests. Due to the poor quality concentration measurements at the surface of the clarifier, the model was not calibrated to perform well in concentration predictions at this surface. However, excellent descriptive capabilities were obtained for sludge profiles and blanket level. The Cho et al. setting velocity function was found to be significantly better in terms of description capability than the more traditional Vesilind function.

Oxygen transferred from the liquid phase into a biofilm can be used for aerobic degradation of organic matter and for nitrification. A second possible pathway for the removal of organic matter is denitrification in anoxic zones deeper in the biofilm. At high organic loading rates with insufficient oxygen supply to the biofilm, denitrification could be induced by providing the biofilm with external nitrate. This possibility was tested in a pilot-scale trickling filter by adding a pulse of nitrate to a highly loaded trickling filter. The experiment showed that denitrification can indeed be induced by adding nitrate at high loading conditions and that this way a considerably increased substrate removal capacity can be obtained. The fact that denitrification occurred was confirmed by the increased production of CO2 from bioconversion processes, without a major change of the O2 consumption. The simplified mixed-culture biofilm model developed by Rauch et al. was extended for the description of off-gas measurements and was able to describe the results of the experiment very well.

The characterisation of the risk of (new) chemicals to species/communities, when both the exposure/environmental concentration and effects (species sensitivity) are variable and uncertain, is the central issue in Probabilistic Environmental Risk Assessment. The spatial variability is one of the largest components of the total variability. This paper tries to explicitly account for this spatial variability by geo-referencing the exposure, effect and finally probabilistic risk. Geo-referencing makes the risk assessment more refined and realistic. In addition, it is also highlighted that geo-referencing the effects of chemicals (species sensitivity distribution) is still a large unexplored area but has large potential to improve probabilistic ecological risk assessments.

An integrated sensor is developed as a tool for monitoring the activated sludge activity on which the performance of the treatment plant depends. The sensor provides information-rich data of high frequency obtained from respirometric-titrimetric and nitrate measurements in one single set-up. The sensor is shown to successfully monitor and provide in-depth insight into nitrification, denitrification and carbon source degradation processes occurring in BNR plants. Based on the experimental results it is hypothesized the ratio of NUR to OUR rather reflects the rate of carbon source uptake (storage) under anoxic and aerobic conditions than growth process.

A (non-exhaustive) survey of new and existing technologies for the monitoring of wastewater treatment plants is presented. Emphasis is given to the way these sensors can provide insight in the ongoing (bio-) processes. Three different uses for sensors can be found: for monitoring (operator support), in automatic control systems and as tools for plant auditing/optimization/modelling by consultants. From this, sensors have been classified in two basic types: (i) reliable, simple and low maintenance sensors for day-to-day monitoring and control and (ii) advanced, higher maintenance sensors that are used in auditing, model calibration and optimisation. The paper is organized according to the typical unit processes of biological wastewater treatment systems: anaerobic digestion, activated sludge, nutrient removal and sedimentation. Attention is drawn to a number of practical problems associated with the use of sophisticated sensors in the harsh (dirty) conditions of wastewater treatment processes. The use of autocalibration and built-in sensor checks, cleaning systems and reliable sample preparation units is illustrated. The paper ends with a discussion of the applicability of the different sensors.

In lab-scale experiments at the 2-stage activated sludge pilot plant of Vienna's central WWTP it is shown that the wastewater soluble COD concentration, which is inert to a sludge with SRT < 1 d (SI(A)) is about double compared to the S(I) concentration in sludge with SRT > 10 d (SI(B)). Unexpectedly the ratio of SI(A)/SI(B) is independent of the sludge age between SRTs of 0.4 and 1.0 days. The difference between the two S(I) fractions is soluble COD that is readily biodegradable by the sludge with SRT > 10 d. However, it is degraded at a lower maximum growth rate. These results comply with earlier results gained with different methods and at different WWTPs. It is hypothesised that very low sludge ages result in a selection of fast growing bacteria, which can utilise only part of the S(S) in the raw wastewater. The other part of S(S) therefore remains in the wastewater and can thus be utilised for enhanced denitrification in the second stage. It is still unknown beyond which sludge age the soluble inert COD SI(A) starts to decrease, finally reaching the value SI(B) for low loaded systems (SRT > 5 days). From this point on S(I) and S(S) are assumed only to depend on the wastewater composition and not on the sludge age. The assumption of the Activated Sludge Model No.1 that the biodegradable fractions can be modelled as a single substrate and by a single removal kinetic (one Monod term) appears not to be applicable for low sludge ages. Some suggestions for mathematical modelling, design and operation of 2-stage activated sludge systems are given.

Partial nitrification techniques, such as the continuously aerated SHARON process, have been denoted for quite a while as very promising for improved sustainability of wastewater treatment. Combination of such a SHARON process with the Anammox process, where ammonium is oxidised with nitrite to nitrogen gas under anoxic conditions, leads to cost-efficient and sustainable nitrogen removal from concentrated streams. In this study practical experiences during start-up and operation of a lab-scale SHARON reactor are discussed. Special attention is given to the start-up in view of possible toxic effects of high ammonium and nitrite concentrations (up to 4000 mgN/l) on the nitrifier population and because the reactor was inoculated with sludge from a SBR reactor operated under completely different conditions. Because of these considerations, the reactor was first operated as a SBR to prevent biomass wash out and to allow the selection of a strong nitrifying population. A month after the inoculation the reactor was switched to normal chemostat operation. As a result the nitrite oxidisers were washed out and only the ammonium oxidisers persisted in the reactor. In this contribution also some practical considerations, such as mixing, evaporation and wall growth, concerning the operation of a continuously aerated SHARON reactor are discussed. These considerations are not trivial, since the reactor will be used for kinetic characterisation and modelling studies. Finally the performance of the SHARON reactor under different conditions is discussed in view of its coupling with an Anammox unit. Full nitrification was proven to be feasible for nitrogen loads up to 1.5 g/l d, indicating the possibility of the SHARON process to treat highly loaded nitrogen streams.

Experimental data are presented that resulted from aerobic batch degradation experiments in activated sludge with simple carbon sources (acetate and dextrose) as substrates. Data collection was done using combined respirometric-titrimetric measurements. The respirometer consists of an open aerated vessel and a closed non-aerated respiration chamber for monitoring the oxygen uptake rate related to substrate degradation. The respirometer is combined with a titrimetric unit that keeps the pH of the activated sludge sample at a constant value by addition of acid and/or base. The experimental data clearly showed that the activated sludge bacteria react with consumption or production of protons during aerobic degradation of the two carbon sources under study. Thus, the cumulative amount of added acid and/or base could serve as a complementary information source on the degradation processes. For acetate, protons were consumed during aerobic degradation, whereas for dextrose protons were produced. For both carbon sources, a linear relationship was found between the amount of carbon source added and the amount of protons consumed (in case of acetate: 0.38 meq/mmol) or produced (in case of dextrose: 1.33 meq/mmol) during substrate degradation. A model taking into account substrate uptake, CO(2) production, and NH(3) uptake for biomass growth is proposed to describe the aerobic degradation of a C(x)H(y)O(z)-type carbon source. Theoretical evaluation of this model for reference parameters showed that the proton effect due to aerobic substrate degradation is a function of the pH of the liquid phase. The proposed model could describe the experimental observations with both carbon sources., (Copyright 2002 Wiley Periodicals, Inc.)

The structural and practical identifiability of a model for description of respirometric-titrimetric data derived from aerobic batch substrate degradation experiments of a C(x)H(y)O(z) carbon source with activated sludge was evaluated. The model processes needed to describe titrimetric data included substrate uptake, CO(2) production, and NH(3) uptake for biomass growth. The structural identifiability was studied using the Taylor series method and a recently proposed generalization method. It showed that combining respirometric and titrimetric data allows structural identifiability of one extra parameter combination, the biomass yield, Y(H), compared to estimation on separate data sets, on condition that the nitrogen fraction in biomass (i(XB)) is known. However, data from short-term batch substrate degradation experiments were not sufficiently informative to allow practical identification of all structurally identifiable parameters. Combining respirometry and titrimetry resulted in improvements of parameter confidence intervals compared to estimation on separate respirometric or titrimetric data sets. However, the level of the improvement seems to be substrate dependent: parameter confidence intervals improved considerably more for dextrose than for acetate degradation models. Noteworthy is the finding that the half-saturation substrate concentrations can be different depending on whether they are estimated from respirometric or titrimetric data. Moreover, this difference appears to be dependent on the carbon source considered: for dextrose, titrimetry-based K(S) values are higher than respirometry-based values while for acetate the opposite was found. It was hypothesized that this can be explained by the different point in cell metabolism where the proton production or consumption takes place, leading to a corresponding difference in timing between pH effect and oxygen consumption. Finally, the biomass yield Y(H) and the nitrogen content of the biomass i(XB) could be estimated from combined respirometric-titrimetric data obtained with addition of a known amount of carbon source. Y(H) can also be estimated from r(O) data when the initial substrate concentration S(S)(0) is known. The values found correspond to values reported in literature, but, interestingly, also seem able to reflect the occurrence of storage processes when pulses of acetate and dextrose are added., (Copyright 2002 Wiley Periodicals, Inc.)

An automatic buffer capacity model building algorithm was developed. The objective of this algorithm is to characterize multiple buffer systems from experimental titration curves. Starting from a simple mathematical model that incorporates the available prior knowledge on the buffer system, the model is sequentially extended by incorporating additional monoprotic buffer equations. To select the most appropriate model from the resulting set of automatically built models, a model structure selection technique was constructed from existing methods.

The objective of this paper was to investigate the usefulness of new multi-criteria measures to evaluate a control strategy through the Benchmark protocol. Using a case study in which respirometry-based control strategies are evaluated, the proposed measures were calculated. An economic index including weighted investment and operating costs (termed Total Cost Index--TCI) appears more powerful than a grey-scale presentation approach. Using the latter approach, it is hard to reflect the relative importance of the criteria investigated, which makes practical decisions rather difficult. In addition, a Robustness Index (RI) is proposed that allows us to evaluate the transferability of control strategies to situations different from the ones defined in the benchmark protocol. Finally, the case study shows that it may be advisable to replace the currently used open loop benchmark reference by a plant in which dissolved oxygen is controlled in all aerated reactors. This quite simple strategy also turned out to be the best one among all evaluated strategies.

The composition of a textile industry wastewater is highly variable, as the industrial process has to follow fashion and season trends. Surfactants represent one of the largest COD fractions in a typical textile wastewater. Therefore, it was the aim of this paper to model the acclimatisation behaviour of an activated sludge system when subjected to composition variations in the surfactant containing feed. The model was based on data obtained in SBR experiments in which a linear alkyl ethoxylate as sole carbon source in the feed was replaced by another with a longer ethoxylate chain. A previously developed model (Fractionated Degradation Model) was applied to each of the 21 SBR cycles carried out in this study. The resulting best-fit parameters were investigated and sub-models were further developed, to create an acclimatisation model, able to predict the sludge acclimatisation level. Using the information given by this model, it was possible to propose an optimal operation scheme to pre-acclimatise the sludge before a surfactant replacement is made in the textile process. A cost analysis was carried out to compare different scenarios, with and without the application of this operation scheme. It was concluded that the proposed pre-acclimatisation process may be cost effective as compared to other scenarios if a cheap surfactant-containing product was employed.

A technique based on laser light diffraction is shown to be successful in collecting on-line experimental data. Time series of floc size distributions (FSD) under different shear rates (G) and calcium additions were collected. The steady state mass mean diameter decreased with increasing shear rate G and increased when calcium additions exceeded 8 mg/l. A so-called population balance model (PBM) was used to describe the experimental data. This kind of model describes both aggregation and breakage through birth and death terms. A discretised PBM was used since analytical solutions of the integro-partial differential equations are non-existing. Despite the complexity of the model, only 2 parameters need to be estimated; the aggregation rate and the breakage rate. The model seems, however, to lack flexibility. Also, the description of the floc size distribution (FSD) in time is not accurate.

The first important issue in pesticide assessment for a watershed is the availability of soil and water quality information in the form of charts, maps or records. However, developing countries generally do not have enough historical data. Thus, sampling programs are crucial for the success of the evaluation, although in developing countries they always represent a controversial task between limited available budget and a precise assessment. This paper shows the steps taken to generate soil and water data to be used in a pesticide assessment project for a 34,000 ha watershed in Ecuador, South America. Sampling campaigns are still being run, so the methodology and the first results are shown here.

An experimental procedure for anoxic activated sludge monitoring with combined nitrate and titrimetric measurements is proposed and evaluated successfully with two known carbon sources, acetate and dextrose. For nitrate measurements an ion-selective nitrate electrode is applied to allow for frequent measurements, and thereby the possibility for detailed determination of the denitrification biokinetics. An internal nitrate electrode calibration is implemented in the experiments to avoid the often-encountered electrode drift problem. It was observed that the best experimental design was with the carbon source in excess, since excess nitrate provoked nitrite build-up thereby complicating the data interpretation. A conceptual model could quantitatively describe the experimental observations and thus link the experimentally measured proton production with the consumption of electron acceptor and carbon source during denitrification.

The merit of activated sludge models depends on the accuracy and reliability of the information they contain on the wastewater to be treated and the biochemical mechanisms involved. In most advanced calibration studies, respirometry i.e. the measurement of the oxygen utilization rate, (OUR), provides the majority of the required experimental database. However, currently used procedures still involve a number of basic and practical problems. Model evaluation of the OUR data may generate a distorted image of the processes involved. Hydrolysis is the most important, yet the most vulnerable process as far as the experimental assessment of accurate kinetic parameters is concerned. This study intends to provide an overview of major experimental limitations in the modeling of activated sludge, with emphasis on the appropriate experimental design for the assessment of the hydrolysis rate.

The status of instrumentation, control and automation (ICA) within the European wastewater community is reviewed and some major incentives and bottlenecks are defined. Future trends of ICA are also discussed. The information is based on a COST 624 workshop and a non-exhaustive survey with regard to ICA carried out in 13 European countries during March 2001. The level of instrumentation (type of sensors, usage frequency, etc.) and how these instruments are used for on-line control purposes are presented for each individual country (Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Netherlands, Romania, Slovenia, Spain, Sweden and Switzerland). The most common types of applied real-time control in wastewater treatment plants are given. One conclusion of the paper is that sensors no longer represent the main bottleneck for on-line control, rather the lack of plant flexibility is more troublesome. Moreover, the current transitional phase of the wastewater industry in Europe represents a unique opportunity to apply ICA on a large scale. The driving forces are simply too strong to ignore.

Uncertainty is a central concept in the decision-making process, especially when dealing with biological systems subject to large natural variations. In the design of activated sludge systems, a conventional approach in dealing with uncertainty is implicitly translating it into above-normal safety factors, which in some cases may even increase the capital investments by an order of magnitude. To obviate this problem, an alternative design approach explicitly incorporating uncertainty is herein proposed. A probabilistic Monte Carlo engine is coupled to deterministic wastewater treatment plant (WWTP) models. The paper provides a description of the approach and a demonstration of the general adequacy of the method. The procedure is examined in an upgrade of a conventional WWTP towards stricter effluent standards on nutrients. The results suggest that the procedure can support the decision-making process under uncertainty conditions and that it can enhance the likelihood of meeting effluent standards without entailing above-normal capital investments. The analysis led to reducing the capital investment by 43%, producing savings of more than 1.2 million euro.

Today, the main concepts required for describing the dynamics of drainage in an entire urban area are known and models are available that can reasonably simulate the behaviour of the urban water system. Still, such integrated modelling is a complex exercise not only due to the sheer size of the model, but also due to the different modelling approaches that reflect the history of the sub-models used and of the purpose they were built for. The paper reviews the state of the art in deterministic modelling, outlines experiences and discusses problems and future developments.

We are witnessing an enormous growth in biological nitrogen removal from wastewater. It presents specific challenges beyond traditional COD (carbon) removal. A possibility for optimised process design is the use of biomass-supporting media. In this paper, attached growth processes (AGP) are evaluated using dynamic simulations. The advantages of these systems that were qualitatively described elsewhere, are validated quantitatively based on a simulation benchmark for activated sludge treatment systems. This simulation benchmark is extended with a biofilm model that allows for fast and accurate simulation of the conversion of different substrates in a biofilm. The economic feasibility of this system is evaluated using the data generated with the benchmark simulations. Capital savings due to volume reduction and reduced sludge production are weighed out against increased aeration costs. In this evaluation, effluent quality is integrated as well.

The urban wastewater system (sewer and treatment plant) has a major impact on the river water quality of urban streams. To minimise this impact, real time control is a valuable option. Since the ultimate goal of any control strategy is to optimise the quality of the river system, it is useful to take pollutant immissions into account when determining the control strategy and/or the setpoints of the controller. However, a simultaneously simulating model of the complete system is needed in order to allow design and evaluation of such control strategies. In this work an integrated model of the urban wastewater system is created. This has been accomplished by implementing surrogate models of the three subsystems within a single software platform. The coupled submodels are subsequently used in a semi-hypothetical case study to optimise the resulting river water quality. An ammonia sensor in the river has been used to control the amount of water treated biologically in the treatment plant. It was shown that this integrated control could lower the peak ammonia concentration in the part of the river downstream of the treatment plant. Hence, a proof of principle has been given that the use of measurements in the river to perform control actions in the sewer system and the treatment plant is a promising option.

Real time control is one of the possibilities to minimise the impact of the integrated urban wastewater system (sewer system and treatment plant) on the receiving water quality. Integrated control uses information about the river state to act in the sewer system or in treatment plant. In order to test and tune these integrated controllers, a simplified integrated model is needed. Even with these simplified models, the simulation times may be too long and further model reduction is needed. In this paper, dependency-structure based model reduction is proposed as a technique to further reduce model complexity. Three steps are proposed: relocation of the upstream system boundaries to just upstream of the first control point, relocation of the downstream boundaries to just downstream of the last measurement point, and third, a further model simplification based on an analysis of the sensitivity of the control actions on submodel elimination. The effect of applying the different reduction approaches on the control strategy and on the resulting river water quality is discussed on the basis of a case study of the catchment of Tielt.

A model is proposed to describe activated sludge acclimatisation to a non-ionic surfactant. The model was calibrated automatically, using WEST, a specific software environment for wastewater treatment model building, simulation and parameter estimation. The assays have been performed in a sequencing-batch reactor (SBR), using a non-ionic surfactant as sole carbon source and non-acclimatised sludge. The best fitting model was based on the assumption of three sequentially degraded COD fractions, where the second fraction is a metabolite of the original molecule and the third fraction is a more slowly biodegradable metabolite resulting from the secondary degradation. For primary degradation, hydrolysis with no associated growth was assumed. The growth of microorganisms responsible for degradation of the second and third COD fractions was presumed to follow Haldane and first order kinetics, respectively. The model was able to fit four consecutive assays of the same acclimatisation process, using Brij 30 as carbon source, with different food/microorganism ratios. The parameters obtained showed that the (self-)inhibition of the growth on the second COD fraction decreased along acclimatisation.

Up to now, within the design/retrofit of wastewater treatment plants (WWTPs), deterministic models were used to evaluate different scenarios on their merits in terms of effluent compliance. This paper describes an approach in which a Monte Carlo engine is coupled to a deterministic treatment plant model, followed by risk interpretation in the form of concentration-duration-frequency (cdf) curves of norm exceedance. The combination of probabilistic modelling techniques with the currently available deterministic models allows to determine the probability of exceeding the effluent limits of a WWTP. This percentage of exceedance is accompanied by confidence intervals resulting from the inherent uncertainty of influent characteristics and model parameters. The approach is illustrated for a hypothetical case study, consisting of a denitrifying plant model inspired by the benchmark model described by Spanjers et al.

A survey was conducted in several Ecuadorian banana plantations to determine the actual pesticide management and their environmental impacts. It was detected that glyphosate, propiconazole, imazalil, tridemorph and imazalil are the pesticides most used in the Ecuadorian banana sector. As a first step, two screening models (EQC and EXAMS) were used to determine the overall distribution of these pesticides in a unitary environment. Whereas EQC evaluates soil, water, sediment and air compartments, EXAMS only takes into account water and sediment compartments. Although both models show different results due to their different approaches, a comparison can still be done considering only the aquatic portion of the unitary environment. For the case study presented here, all 5 pesticides tend to affect more the soil and water compartment.

A sensor based on on-line measurement of nitrate and acid/base titration is proposed for anoxic activated sludge monitoring. The data probed from the preliminary experiments with denitrifying activated sludge samples using anoxic-titrimetric sensor are information-rich with high data frequency that can be used further in model-based data interpretation and parameter estimation of denitrification process. Moreover, the sensor was able to detect intermediate accumulation, which is an often-observed phenomenon during the denitrification process. It therefore offers the possibility for use in overall performance evaluation of biological nutrient removal plants.

In this paper, buffer capacity profiles are used in the framework of automatic monitoring of water quality. The aim of the proposed methodology is to automatically and stepwise build buffer capacity models for each particular titrated sample, and to quantify the individual buffer systems that constitute the total buffer capacity. An automatic and robust model building algorithm has been developed and applied to many titration curves of effluent and river water samples. It is illustrated that the application of automatically built buffer capacity models mostly results in similar or better estimations of ammonium and ortho-phosphate in the samples compared to a priori fixed buffer capacity models. The automatic modelling approach is also advantageous for alarm generating purposes on e.g. river waters, because unexpected buffers are easily detected.

Even though many models have been proposed for primary clarification, none is directly compatible with the ASM1. The objective of this paper is to present the development of a reactive primary clarifier model to be used in a wastewater treatment plant simulator (WEST). A model simulating COD behavior has been developed based on the Takacs model, and was tested with full-scale data. Particulate effluent COD was well described but problems occurred predicting the underflow suspended solids concentration. The model had to be upgraded with a residence time and a flocculation term to simulate the behavior of soluble COD. An ammonification term was added to the model, resulting in an improved model fit on effluent ammonium.

An earlier study on theoretical identifiability of parameters for a two-step nitrification model showed that a unique estimation of the yield YA1 is possible with combined respirometric-titrimetric data, contrary to the case where only one type of measurement is available. Here, the practical identifiability of model parameters was investigated via evaluation of the output sensitivity functions and the corresponding Fisher Information Matrix (FIM). It appeared that the FIM was not sufficiently powerful to predict the practical identifiability of this case with combined measurements as parameters could indeed be identified despite the fact that the FIM became singular. The accuracy of parameter estimates based on respirometric and titrimetric data and combination thereof was also investigated. Estimation on titrimetric data (Hp) was very accurate and a fast convergence of the objective function towards a minimum was obtained. The latter also holds for estimation on oxygen uptake rate data (rO), however with a lower accuracy. Parameter estimation based on oxygen concentration data (SO) was more complex but resulted in a higher accuracy. Thus, when the highest accuracy is needed it is recommended to estimate parameters initially on Hp and/or rO data, and to subsequently use these parameters as initial values for final, and more accurate estimation on SO data.

In this paper, the practical implementation and validation of advanced control strategies, designed using model based techniques, at an industrial wastewater treatment plant is demonstrated. The plant under study is treating the wastewater of a large pharmaceutical production facility. The process characteristics of the wastewater treatment were quantified by means of tracer tests, intensive measurement campaigns and the use of on-line sensors. In parallel, a dynamical model of the complete wastewater plant was developed according to the specific kinetic characteristics of the sludge and the highly varying composition of the industrial wastewater. Based on real-time data and dynamic models, control strategies for the equalisation system, the polymer dosing and phosphorus addition were established. The control strategies are being integrated in the existing SCADA system combining traditional PLC technology with robust PC based control calculations. The use of intelligent control in wastewater treatment offers a wide spectrum of possibilities to upgrade existing plants, to increase the capacity of the plant and to eliminate peaks. This can result in a more stable and secure overall performance and, finally, in cost savings. The use of on-line sensors has a potential not only for monitoring concentrations, but also for manipulating flows and concentrations. This way the performance of the plant can be secured.