Analyzer

The Analyzer module allows you to automatically perform steady-state or dynamic sensitivity analyses on model parameters, which will save you a significant amount of time and effort. The feature is available in the "Pro" configuration of GPS-X ^TM

Examples of what can you achieve with the Analyzer

Steps

GPS-X ^TM performs successive steady-state simulations for a range of wastewater temperatures and plots the effluent ammonia as it is calculated.

Solution

Figure 1 shows effluent ammonia (Y-axis) as a function of wastewater temperature (X-axis). From this graph, you can determine that some nitrification is begins to occur at around 7℃, and full nitrification is established at 20℃ and above. Determine the sensitivity of a nitrifying plant to wastewater temperature while experiencing a sinusoidal influent flow rate.

Steps

GPS-X ^TM runs a five-day dynamic simulation with a sinusoidal influent flow rate. This five-day run is repeated for a number of different wastewater temperatures (three in this example), and the results are plotted on the same output graph, as shown in Figure 2. In this type of analysis, each line on the graph corresponds to the level of ammonia in the final effluent at a different wastewater temperature.

Solution

As displayed in Figure 2 there is increased nitrification occurring at higher influent temperatures. The concentration of effluent ammonia varies with time due to the sinusoidal influent flow rate.

Influent or sidestream (supernatant, etc.) flow rate or loading

Tank sizes

Storm flow management (bypass, storage)

Wasting, recycle, air flow and chemical dose rate

Automatic controller setpoints and tuning parameters

Kinetic and stoichiometric parameters

Settling parameters (e.g. SVI)

Observe the effect of plant performance measures, such as:

Effluent concentrations

Nutrient removal efficiency

Pollutant load to receiving waters

Plant operating variables (e.g. MLSS, DO)

Sludge production

Monte Carlo analysis in GPS-X ^TM allows users to generate probabilistic distribution of model outputs based on inputs with predefined probabilistic distribution. Both steady-state or dynamic Monte Carlo simulations can be performed in GPS-X ^TM . This analysis can be used to assess the effect of uncertainty in model parameter and or model input on the model outputs. A robust Monte Carlo Analysis Tool allows users to evaluate plant performance when not all of the model inputs are well-known. Users can now run simulations using a typical distribution of values, rather than being forced to choose a single value, and observe a range of model outputs. For example, users can predict a range of effluent ammonia concentrations given the typical range of nitrifier growth rates, even though the true actual growth rate isn't known. The Monte Carlo Analysis Tool is a valuable addition to the GPS-X ^TM toolkit, and provides users with a way to bring the uncertainty of real life into their models.

Optimizer

The Optimizer is designed to solve parameter estimation and process optimization problems. The feature is available in the "Pro" configuration of GPS-X ^TM

The objective is to determine the values of the heterotrophic maximum specific growth rate and the readily biodegradable substrate half saturation coefficient that allow for optimal matching to the diamond-shaped points on the graph displayed in Figure 1. These points are the results from a lab-scale bench experiment.

Steps

In this example the GPS-X ^TM optimizer is searching for the coefficient values that will minimize the difference between the simulated results and actual lab data. The small diamond-shaped points represent the measured filtered COD, while the continuous lines are the simulated results of the batch experiment as GPS-X ^TM repeatedly searches for the appropriate coefficient values.

Figure 1. Kinetic parameter optimization showing successive guesses getting closer to the measured data points as the GPS-X ^TM optimizer automatically calibrates the model.

Solution

The optimized output is presented in Figure 2.

A similar calibration exercise using the optimizer can also be based on measurements from a full-scale wastewater treatment facility. For example, the optimizer could be used to find nitrifier growth rates using measured effluent ammonia data. The GPS-X ^TM Optimizer can be used to find appropriate control methods to minimize effluent concentrations from an activated sludge process.

Steps

The plant shown in Figure 3 is designed for biological nutrient removal. Under current operating conditions, this plant has an effluent total nitrogen (TN) of 10.0 mg/L. In this example, the optimizer was used to vary the RAS flow, the WAS flow and the mixed-liquor internal recycle (MLIR) flow to meet a more stringent effluent TN objective of 7.0 mg/L. The objective of this optimization is to minimize the effluent TN concentration by finding the best combination of the three flows, while ensuring that the maximum installed pumping capacities at the plant are not exceeded. This optimization required 70 iterations to converge.

Solution

The table in Figure 4 shows the results of the optimization. The iteration number and the three optimized parameter values are shown along with the associated effluent TN concentration. It can be seen from the results of this optimization that the plant can meet an effluent TN objective of 7.0 mg/L, if the current operating conditions are modified to the values determined by the GPS-X ^TM Optimizer.

Automated model calibration

Use the optimization module to find model parameter values that provide the best fit between measured and simulated data.

Multi-parameter optimization ensures that when multiple parameters impact the same variable, the best set of parameter values can be determined.

Evaluate Design and Process Control Options

Use for plant optimization by finding new control option combinations to maximize the use of existing plant infrastructure.

Determine the best trade-off between multiple process design options.

Optimize SBR cycle settings to maximize biological nutrient removal.

Find the most cost effective operating strategy by using the optimizer to minimize total annual operating cost.

It can be used with on-line data or on a set of off-line time series data. For details on using on-line data, contact us. The feature is available in the "Pro+" configuration of GPS-X ^TM

The Approach

In GPS-X ^TM , dynamic parameter estimation is done by applying the time series optimization approach mentioned earlier to a moving time window. Instead of estimating parameters from an entire set of data, GPS-X ^TM calculates a set of parameter estimates for each time window using the parameter estimates from the previous time window as a starting guess. This approach can be used on a data file that is continually updated with new blocks of data or on a static file of time series data. Note that you can use any of the objective functions that are available for time series optimization when doing dynamic parameter estimation. The length of the time window controls how often the parameters are updated. The shorter the time window, the more often the parameters are updated. To improve the fit between the model results and data, a model parameter can be varied over the simulation period.

Steps

Typically the clarifier's flocculent zone settling parameter is set to one specific value over an entire calibration time period; this result is presented in Figure 1. Holding this parameter constant does not result in the best fitting of the data, so this parameter can be varied with time to improve the agreement between the simulated model results and points on the graph, and to indicate the dynamic response of this parameter. As presented in Figure 2, the flocculent zone settling parameter in the primary clarifier is adjusted at each 0.4-day time step to improve the fit of the model and actual data for the primary effluent TSS concentration presented in Figure 3.

Solution

Through comparison of Figure 1 and Figure 3, dynamically adjusting the flocculent zone settling parameter provides improved fitting to the data.

Online

The GPS-X ^TM on-line tool is a very powerful tool that allows users to continually add in new blocks of data rather then using a static input file. The feature is available in the "Pro+" configuration of GPS-X ^TM In on-line mode, these input files are growing continuously, line by line, and GPS-X ^TM monitors and reads in the data at a specified sampling rate. If a new line appears in the data file, it will be brought into GPS-X ^TM (the file input controller updated, or the data appears as a new point on the graph) and the simulation advances to the new time. Thus, the simulation would be continuously updated with data arriving in real time from a plant SCADA system into the input data file, and real-time simulation of the plant would be achieved. This allows you to see a data/model difference from the very moment it starts to emerge, indicating a problem. An additional tool called "sigtrack", found in the toolbox object, could be used to set up alarms if the data, simulation, or the error between them goes beyond certain limits. An advanced on-line application (which would require the Dynamic Parameter Estimator, DPE) would do a real time simulation and calibration/optimization simultaneously. In this way the model calibration would be automated using the most up to date plant data.

Customizability

Open Modelling Platform with Open-Source Models

The GPS-X ^TM simulation environment allows users to write, edit, or change models, via the ACSL programming language. Users can add any new calculations, algorithms, or variables to their model layout.

Advanced tools such as Model Developer allow for easy biological model manipulation in matrix format.

All GPS-X ^TM input and output menus, as well as the new simulation result summaries, can be edited and customized by the user.

Wastewater plants are variable, and it can be difficult to create a simulation tool that can handle all possible layouts and desired variables. Taking advantage of the customizability available in GPS-X ^TM users can generate input and output variables and use them as if they were part of the original layout.

Steps

Figure 1 presents a graph of the influent flow and the oxygen uptake rate. New input and output variables are created by the user using the ACLS language. GPS-X ^TM incorporates these variables into the user-defined input and output menus, so in this particular instance they can be used as input controllers or added to an output graph.

Solution

Figure 2 presents the same graph as in Figure 1 but with the addition of the user defined variable for the oxygen uptake rate with added noise.

Figure 1. Influent Flow Rate and Actual Oxygen Uptake Rate vs. Time.

Figure 2. Influent Flow Rate, Actual Oxygen Uptake Rate, and Oxygen Uptake Rate With Noise vs. Time.

Fit Other Components into GPS-X ^TM

We have developed an integrated interface for creating and executing python scripts from within GPS-X ^TM . The python script can interact with GPS-X ^TM models to exchange data and allow you to incorporate supplemental analytics on the model data. By allowing custom scripts to be run in GPS-X ^TM , you can use all of the tools available in python to extend what is possible in GPS-X ^TM . While the applications of GPS-X ^TM are diverse, it may not always be built generally enough to provide the exact analysis or data visualization options you would like for your project. By using python, you can do things such as create custom plots, introduce new types of statistical analysis and automate repetitive tasks in ways that are not possible using only GPS-X ^TM .

Fit GPS-X ^TM into Your System

While you can use python to fit a 3rd party component into GPS-X ^TM , you can also use python to integrate the GPS-X ^TM model as a component of your bigger enterprise software. Using the GPS-X ^TM open API and python, you can programmatically start, communicate and extract results from a model directly from your own software solutions. GPS-X ^TM may be the puzzle piece that you've been searching for.

MATLAB

The MATLAB Link allows the user to call MATLAB code from a GPS-X ^TM model, sending GPS-X ^TM model variables as inputs to a special function M-file, and mapping the outputs of that function to other GPS-X ^TM model variables. In effect, MATLAB is placed "in the loop" during a simulation, as depicted in the diagram. This makes it possible to use the MATLAB Link to control a GPS-X ^TM model using a controller that is implemented in MATLAB, thereby leveraging MATLAB's extensive library of control-related functions. The strength of GPS-X ^TM lies in its ability to rapidly model and simulate wastewater processes. Controllers are an important part of the plant, and basic PID controllers are built into GPS-X ^TM - the user sets the controller parameters in the appropriate pop-up windows. The best-performing control systems are designed, rather than tuned heuristically. As described above, the tools in MATLAB's Control System Toolbox provide excellent controller design capabilities. Therefore, by providing an interface between GPS-X ^TM and MATLAB, you will be able to model wastewater processes in GPS-X ^TM , design suitable controllers for these processes, and evaluate the performance of control systems once the controllers have been transferred back to GPS-X ^TM . Other possible applications of MATLAB with GPS-X ^TM include data logging, post-run analysis of GPS-X ^TM simulation results, and displaying data using MATLAB's three-dimensional graphical capabilities.

*MATLAB is a registered trademark of MathWorks

Model Developer

Model Developer is a tool integrated into GPS-X ^TM that allows you to create new biological models by editing the model matrix quickly. The feature is available in the "Pro" configuration of GPS-X ^TM

Model Developer allows you to :

Add New

Add new rates (rows in the matrix) or involve additional states (columns in the matrix)

Change Existing

Change kinetic equations or stoichiometric relationships

Modify Values

Modify default values of various parameters