Content uploaded by Lloyd R Townley
Author content
All content in this area was uploaded by Lloyd R Townley on Jan 27, 2020
Content may be subject to copyright.
INTRODUCTION
WRRM1 and WRRM2: Implementations in GoldSim
of Unit Process Models and IWA Benchmark Models
(BSM1 and BSM2) for Nutrient Removal
Townley, Lloyd R.*, Jiang, Huanhuan* and Tang, Jinquan*
*Nanjing University Yixing Environmental Research Institute, Yixing, Jiangsu, China, townley@njuyi.cn
New simulation software has been developed at the Nanjing University Yixing Environmental Research Institute, for use in training, and to facilitate the use of simulation software for improving the
performance of water resource recovery facilities (WRRFs) in China. Computer simulation of WRRFs (formerly known as wastewater treatment plants or WTTPs) is useful for:
1. design of new WRRFS,
2. design of modifications to existing WRRFs,
3. diagnostic analysis to understand the behaviour of existing WRRFs,
4. design of new control strategies to improve the performance of WRRFs, and
5. training of operators, professionals and students.
The new implementations are the first complete implementations in many years. This is also the first time that GoldSim has been used to simulate WRRFs.
The main reason for developing new software was the desire to have a standalone executable that can be launched by calls from an external control program. This will allow development and testing
of new algorithms based on many simulations being run in parallel.
The BSM2 flowsheet has a primary
clarifier, 5 ASM1 activated sludge
reactors, a Tak ács secondary
clarifier, a thickener, an ADM1
anaerobic digestor, dewatering and
a storage tank that is bypassed. All
details of BSM2 have been
implemented in WRRM2, except
the analysis of risk of filamentous
bulking, based on fuzzy logic.
There are many options for viewing
the model results while it is running,
including animation of TSS and
COD in all parts of the facility.
Our efforts would not have been possible without the efforts of dozens of experts in the past
30 years, culminating in the development and publication of BSM1 and BSM2 under the
guidance of the IWA Task Group on Benchmarking of Control Strategies for Wastewater
Treatment Plants.
We chose to use a graphical software development environment known as GoldSim [5],
largely because of the first author’s experience using GoldSim for integrated water
management in the mining sector. This is the first time that GoldSim has been used for
simulating the conventional activated sludge (CAS) process, let alone anaerobic digestion or
a Takács secondary clarifier. GoldSim always steps forwards in time, so it is naturally suited
for explicit soutions of Monod-type equations. A very useful feature of GoldSim is the fact
that units are automatically checked and converted. GoldSim can read data from and write
results to Excel files; it can call embedded models saved as dynamic linked libraries (DLLs).
Most importantly, a free GoldSim Player can be used to execute GoldSim models, saved as
.gsp files, so that many models can be executed in parallel.
METHODS EXAMPLE OUTPUT
inspiring change
Please ask us to demonstrate WRRM1 and WRRM2, and ask how you could use the software
yourself for teaching students.
Acknowledgements: The authors are grateful for financial support from the Nanjing University Yixing Environmental Research Institute and ongoing assistance from the Environmental Science and Technology Park (ES&TP) in Yixing.
The first author is grateful to the Jiangsu Industrial Technology Research Institute (JITRI) for support as aJITRI Researcher.We are grateful to Prof.David Waite of the University of New South Wales (UNSW) for
supporting the first author as aSenior Visiting Fellow, to Rick Kossik and his team at GoldSim Technology Group for their support, to Prof.Ulf Jeppsson of Lund University in Sweden for his continued support of the
MATLAB versions of the IWA benchmark simulation models and to Dr Xavier Flores-Alsina of the Technical University of Denmark (DTU) for his assistance and encouragement..
References: [1] Alex, J., Benedetti, L., Copp, J., Gernaey, K.V., Jeppsson, U., Nopens, I., Pons, M.N., Steyer, J.P. & Vanrolleghem, P. (2008a) Benchmark Simulation Model no.1 (BSM1). IWA Report.
[2] Alex, J., Benedetti, L., Copp, J., Gernaey, K.V., Jeppsson, U., Nopens, I., Pons, M.N., Rosen, C., Steyer, J.P. & Vanrolleghem, (2008b) Benchmark Simulation Model no.2 (BSM2). IWA Report.
[3] Copp, J.B., Jeppsson, U. & Vanrolleghem, P.A. (2008) The Benchmark Simulation Models –A Valuable Collection of Modelling Tools. Proceedings, iEMSs 2008: International Congress on Environmental Modelling and
Software, International Environmental Modelling and Software Society.
[4] Gernaey, K.V., Jeppsson, U., Vanrolleghem, P.A. & Copp, J.B. (2014) Benchmarking of Control Strategies for Wastewater Treatment Plants. IWA Publishing, London.
[5] GoldSim Technology Group (2018) GoldSim User’s Guide, Version 12.1, 1191 pp. See www.goldsim.com.
RESULTS and CONCLUSION
BSM1 and BSM2 have been implemented in GoldSim. The new implementations are known
as “water resource recovery models”, WRRM1 and WRRM2, respectively. Each model exists
in two forms.
WRRM1-Validation and WRRM2-Validation are identical to the MATLAB benchmarks. The
only detail not implemented in WRRM2-Validation is the analysis of risk of filamentous
bulking, based on fuzzy logic. Each model includes influent data identical to that used in the
benchmarks. The benchmarks were run using MATLAB, and the results were imported and
saved within WRRM1-Validation and WRRM2-Validation; this allows us to demonstrate that
the new models produce identical results to the MATLAB benchmarks, to several significant
figures. WRRM1-Validation and WRRM2-Validation have many graphical screens and allow
users to explore the rates of biochemical reactions, with a level of detail that is not possible in
MATLAB without considerable additional programming.
WRRM1-Control and WRRM2-Control allow users to modify control handles, to attempt to
improve the performance of the benchmark facilities using standard influent data. After
completing the implementation of WRRM1, the authors learned that the developers of BSM1
intended the benchmark facility to be overloaded (undersized). As a result, it is difficult to
improve the performance of the facility. For this reason, WRRM1-Control allows users to
modify the sizes of activated sludge reactors, and other components of the facility, thereby
providing more degrees of freedom in the model. Similarly, WRRM2-Control allows more
degrees of freedom than BSM2, although it still allows only the standard influent data.
Rigorous comparison with the original benchmarks has allowed the developers to have
confidence in their ability to implement the equations representing biochemical processes.
The next phase of development of WRRM1 and WRRM2 will focus on customization of
models for specific plants. The software can be modified to simulate any number of activated
sludge reactors, in common configurations such as A2/O, A/O, oxidation ditch etc. Current
efforts are focused on developing methods to estimate the influent characteristics based on
typical measured data.
We now have confidence in our ability to customise models for individual WRRFs in China,
with the same or similar unit processes as in BSM1 and BSM2.
www.iwahq.org
Table 1 –Key references that describe the development of BSM1 and BSM2
Reference Comments
Alex et al. (2008a) [1] Report on development of BSM1. Revised in 2018. See
www.iwa-mia.org/benchmarking/.
Alex et al. (2008b) [2] Report on development of BSM2. Revised in 2018. See
www.iwa-mia.org/benchmarking/.
Copp et al. (2008) [3] Conference paper at time of release of benchmark models.
Gernaey et al. (2014) [4] Final report of the IWA Task Group on Benchmarking of Control
Strategies for Wastewater Treatment Plants. Contains a list of
15 technical reports of the Task Group, not all of which have
been finalised.
WRRM1-Validation
GoldSim is a graphical software
development environment; the
equations are hidden inside
WRRM1 and WRRM2, but they
have been rigorously tested
against MATLAB results.
WRRM1-Validation shows perfect
agreement with MATLAB results,
shown here in the 5th ASM1
reactor.
green lines
plot on top of
red lines
WRRM2
