Tel: +352 24 83 69 76    Twitter Icon  
RTC4Water - October 2019

The benefits of a fully autonomous, real-time control software used to optimize flow in mixed sewer networks – with the primary benefit being a reduction in combined sewer overflow events – are well understood and accepted1,2,3. But what cost efficiencies could be gained if, during the design phase of a wastewater treatment plant, engineers incorporated the use of this type of optimization software into their overall performance planning? What cost savings could be realized by reducing basin size, pumping or pipe capacity requirements while still achieving the same performance levels imposed by regulatory standards?

 

A Real-World Study Using Proven Software Tools


In 2012 the research team of what would soon become RTC4Water developed and installed Luxembourg’s first fully autonomous, predictive sewer network optimization software for a large, rural wastewater treatment plant. This installation has provided both the company and the treatment plant operator with a wealth of information and has become a proven platform for effectively evaluating new ideas. Recently, Luxembourg’s national water agency, l’Administration de la Gestion de l’Eau or AGE, partnered with RTC4Water to assess how classically developed engineering plans might be improved if the sewer network utilized a real-time network optimization software – in this case RTC4Water’s Global Predictive Controller (or GPC). For simplicity, this article will refer to a sewer network that uses this type of software as one that is dynamically managed (DM). The goal of the AGE study was to determine if engineering companies could reduce certain costs – financial as well as ecological – while still meeting required engineering norms and standards. As the wastewater treatment plant mentioned above was about to undergo an expansion project, a tangible, real-world assessment could be rapidly developed to find answers to these questions.

 

The Scope and Scale of the Study


The AGE supplied RTC4Water with the engineering plans for the new construction project. The area under consideration consisted of five rural communities which had been divided into 20 catchment areas. The design plans called for five mixed rainwater/sewage overflow (CSO) basins which were connected to a network and which then terminated at an existing wastewater treatment plant (WWTP). Rain fall data was provided by the AGE and another planning office developed the pollution load calculations (using KOSIM) which were taken into account by the engineering office to develop the final “classical static management design”, or CSM proposal (which also incorporated the German DWA A128 norm for pollution loads). To begin the study, the research team first focused on CSO outflow rates and overflow frequencies. Some of the scenarios which were developed for this initial analysis were: an assessment of a dynamically managed network against the CSM network whereby the maximum outflow of the CSO basins was not adjusted, a dynamically managed network compared against the CSM network whereby an adjustment of the CSO outflow was made (increased) but no modifications made to the intake flow rate of the WWTP, and finally an assessment of a dynamically managed network against the CSM network whereby an adjustment of the CSO outflow was made (increased) and intake flow into the WWTP was increased for a limited period of time. The research team also explored the scenario of a dynamically managed network against the CSM network whereby an adjustment of the CSO outflow rate was made (increased), a different pollution load classification for one of the CSOs in the network was added, but again no modifications made to the intake flow rate to the WWTP.

 

Virtual Environments Allow for Rapid Analysis


Naturally, before beginning the assessment of a dynamically managed flow approach, it is necessary to first have a virtual model which reflects the characteristics and behavior of the proposed CSM network design. As RTC4Water’s GPC software interacts with SWMM in a very efficient way, this software tool was used to develop the initial simulation environment for the non-dynamically managed network. The GPC optimization (dynamic management) software works with SWMM by querying the variables recorded in the SWMM model at every stage of the simulation, calculates the best control set points for a specific point in time, and then send these results back to the SWMM software with the data then being used to set the parameters for the next stage of the SWMM simulation. To represent the virtual operations of the dynamically managed network design, a relatively simple model of the proposed network was developed within the GPC platform. In this way, a second virtual simulation environment was established which allowed different optimization scenarios to be explored at a very rapid rate. Inputs to SWMM-simulated rainwater catchment basins were used, which were then dynamically managed within the GPC virtual environment which would apply a fully autonomous, predictive flow control approach. At each step of the simulation (preformed every 10 minutes), the GPC stored not only the optimization results, but also the optimization problem to be solved and any network restrictions that might limit the final solution. Analysis of this data showed which changes to the network’s infrastructure design parameters (e.g. flow rates, basins sizes) would offer the greatest potential for improving overall network performance.

 

Results 


For illustration purposes, the research team have selected data from a week in February 2017 whereby the proposed non-dynamically managed - or CSM designed - network would have encountered medium sized overflow events. To keep this article short and concise, only a sampling of these results will be shown. If you are interested in a full description of the study parameters and the detailed results for each scenario, please This email address is being protected from spambots. You need JavaScript enabled to view it. RTC4Water and they will make the paper available to you.

 
Scenario 4.2: comparison of a dynamically managed network (against a classical static network, or CSM) whereby the basin outflow was increased but no modifications made to the intake flow rate for the WWTP.


In this scenario, the maximum output for each of the five CSOs in the proposed network was increased, but no modification was made to the nominal input rate of the WWTP. In this example, a dynamically managed system would have more leeway to exploit the full dynamics of the sewage system and thus realize a further reduction in CSO events. For gravity outlets and larger rainwater retention basins, a significant increase in outflow was specified (Table 2). In this scenario, use of a DM approach would have resulted in a very significant 22% reduction in CSO events as compared to a CSM approach (Ref. 4.2, Table 1 below).

Table 1: CSO Events of a DM System vs. Classical or Static Management Approach

CSO Events DM vs Classical Static Design

Table 2: CSO Flow Rates for Scenario 4.2

CSO Basin Name Qab (l / s) QMax (l / s)
RUB01 4.0 4.5
RUB02 16.0 25.0
RUB03 5.4 9.0
RUB04 9.8 17.0
RUB05 2.3 4.0

 

As an additional test, in this same scenario the receiving waters for one of the CSO basins (named RUB03) was classified as being more polluted and should therefore be managed differently from other basins in the network. For example, all CSO basins in the GPC virtual model were given a sensitivity parameter of 1.0 with the exception of RUB03 which was set to a value of 1.3. Table 1 (Ref. 4.2a above) shows that a dynamic management approach will displace some of the overflow volume from RUB03 to RUB02, without having to accept a significant loss of overflow performance.


Scenario 4.3: an assessment of a dynamically controlled network (again, against a classically designed, static network or CSM) whereby the basin outflow was adjusted and intake flow into the WWTP was increased for a limited period of time. 


One of the advantages of RTC4Water’s GPC software is that it automatically and continuously delivers a more consistent flow into the WWTP. Automatically initiating short-term flow increases - provided that this does not exceed the maximum input parameters of the treatment plant - has a significant impact on the sewer system’s performance. In this scenario, the nominal WWTP flow rate went from 32 (l / s) to a maximum of 34 (l / s) for a short period of time. By using a DM software like the GPC, this flow rate adjustment can be automatically made if an overflow event has been forecasted by the software. In this scenario, the prediction horizon of the GPC was increased from 50 [min] to 100 [min] to detect potential overflows earlier and to then automatically make adjustments. Table 1, (Ref. 4.3, above) shows that in this situation a CSM design will result in a 30% greater overflow volume when compared to a DM approach. Figure 1 below shows that at the beginning of the event, the flow entering the WWTP is set to a maximum of 34 (l / s) for about 6 hours, but is then automatically reduced to a nominal flow (where KA represents the WWTP).

Figure 1: CSO basin volumes (left) / basin outlet (right) with temp. increased inflow to WWTP

 

Conclusions


While only a first snapshot from the overall study, this analysis does demonstrate that the utilization of a dynamic management software can impact the approach to sewer network design. Another way of looking at this is to say that engineering companies and national regulators can implement more environmentally friendly and cost-effective network designs without sacrificing performance or infringing on norms or standards. In another example, if we use actual operational costs for the network in this study, we can see that an increase in pumping power from 5.4 (l / s) to 9 (l / s) corresponds to an increase in pumping costs of 7.774,00 €. But by incorporating a dynamic management approach (while still respecting standards – the DWA A128 pollution guidelines) a reduction in the flow volume from the CSOs (for example reducing RUB03 from 179 m³ to 174 m³) cancels out this additional cost. Yet this same volume reduction would deliver a 20.9% savings of the total overflow volume. And by further adjusting the sensitivity level of basin RUB03, overflow could even be further reduced - in this case, to zero - for the given rain event.
As we continue our work with the AGE, we hope to present more information from this study. We would very much like to hear your comments or help you with any questions you might have. Please feel free to send an This email address is being protected from spambots. You need JavaScript enabled to view it. to the team to learn more about this project.

 

For more information, please see these reference papers on the use of Real-Time Control (Dynamic Managment) software:

1. https://www.mdpi.com/2073-4441/10/11/1675

2. https://pdfs.semanticscholar.org/a0dc/a13f1aec89e5366d2ec955a4e2dbce4f8cce.pdf

3. https://iwaponline.com/wst/article/2017/2/552/38794/Astlingen-a-benchmark-for-real-time-control-RTC

 

 

 

 

 

Published in RTC4Water Blog