BMS2-Python user guide¶

This guide provides an overview and explains how to set up your own BSM2-Python project.

What is BSM2-Python?¶

BSM2-Python is an implementation of a wastewater treatment plant using the Benchmark Simulation Model No. 2, which is extended with an energy management system that operates the plant's energy processes. BSM2-Python can be used to fully display a real-world wastewater treatment plant to simulate and run techno-economic analyses. It is possible to modify the predefined BSM2-Python layout by adding, removing or modifying units of BSM2-Python to build a customized plant layout.

The following figure illustrates the flowchart for wastewater treatment and energy management. Initially, wastewater flows into the primary clarifier, where excess sewage sludge is removed. Next, the wastewater flow undergoes treatment in five activated sludge reactors. The treated effluent is then fed into a settler, where any remaining sewage sludge is removed. The sludge collected in the settler is compressed in a thickener, separating excess wastewater, which is returned to the treatment process. In the anaerobic digester, sewage gas is produced and subsequently stored in a biogas storage. The fully fermented sewage sludge is drained in the dewatering unit and then disposed of. Excess wastewater is held in a wastewater storage tank and recycled back to the wastewater treatment processes. In the energy management system, the sewage gas is utilized in two combined heat and power (CHP) units to generate electricity for the aeration in the activated sludge reactors and heat for the anaerobic digester. If the heat generated by the CHP units is insufficient, additional sewage gas is burned in a boiler. When sewage gas production exceeds the storage capacity, the surplus is burned off with a flare. To regulate the heat network, a cooler is used to dissipate any excess heat.

BSM2-Python components¶

Further details about the components used for wastewater treatment and energy management are provided in the section below.

Wastewater treatment components¶

Energy management components¶

Quickstart¶

Run default model¶

You can use the following convenience function:

from bsm2_python import BSM2OL

# initialize the BSM2 Open Loop model
bsm2_ol = BSM2OL()

# run the simulation
bsm2_ol.simulate()

This will run the BSM2 Open Loop model for the default 609 days of simulation time. It will then plot IQI, EQI and OCI values for the effluent over the last few days of simulation. Additionally, relevant data will be saved to data/output_evaluation.csv for further analysis.

Run with custom aeration¶

You can also run the BSM2 models with your own aeration control. This example selects a random kla value for each reactor every timestep. The final performance is saved in the oci variable:

import numpy as np
from bsm2_python import BSM2OL
from tqdm import tqdm


bsm2_ol = BSM2OL()

# The kla values to choose from
select_klas = np.array([0, 60, 120, 180, 240])

for idx, _ in enumerate(tqdm(bsm2_ol.simtime)):
    # select random klas for all five ASM1 reactors
    klas = np.random.choice(select_klas, 5)

    # make a step in the simulation with the specified kla values
    bsm2_ol.step(idx, klas)


oci = bsm2_ol.get_final_performance()[-1]

Run Closed Loop simulation with custom DO setpoint¶

You can also run the BSM2 Closed Loop model with your own dissolved oxygen (DO) setpoints. Please note: The Closed Loop model runs with a resolution of 1 minute to ensure sensor stability. As a result, the simulation might take a while to complete.

from bsm2_python import BSM2CL
from tqdm import tqdm


bsm2_cl = BSM2CL()

# The custom DO setpoint for the BSM2 default aeration control
so4_ref = 1.5

for idx, _ in enumerate(tqdm(bsm2_cl.simtime)):
    bsm2_cl.step(idx, so4_ref)

# get the final performance of the plant
oci = bsm2_cl.get_final_performance()[-1]

Run with energy management model¶

We introduced a simple energy management model (including CHPs, Boilers, Flares and a small techno-economic analysis) that can be used to simulate the energy consumption and production of the plant.

from bsm2_python import BSM2OLEM

bsm2_olem = BSM2OLEM()

bsm2_olem.simulate()

# get the cumulated cash flow of the plant
cash_flow = bsm2_olem.economics.cum_cash_flow

Short documentation of the simulation¶

A basic simulation (with energy management) in this project always consists of the following steps:

1. Initialization

2. Simulation loop

   • Stabilization of wastewater treatment plant (WWTP)

   • Simulate WWTP

   • Collect data from WWTP

   • Simulate energy management (EM)

   • Collect data from EM

3. Evaluation

A detailed documentation of the BSM2-base module can be found below:

Build a customized BSM2-Python plant layout¶

Both the WWTP layout as well as the energy management layout can be customized by adding, removing or modifying individual process units. To customize the WWTP layout the module bsm2_base.py has to be changed. For the customization of the energy management layout change the module bsm2_olem.py.

Customizing the wastewater treatment plant layout¶

To showcase how to customize the WWTP layout, a real-world WWTP with population equivalent (PE) of 435,000 was re-created with BSM2-Python from the case study Plant-wide modelling for assessment and optimization of upgraded full-scale wastewater treatment plant performance ¹. The WWTP layout from the study consists of a mechanical treatment (screens, grit and grease chamber), a biological stage with a suspended biomass activated sludge process (three parallel plug-flow aerobic reactors and four parallel secondary clarifiers) and sludge treatment (sludge thickening, anaerobic digestion, dewatering and sludge drying).

The WWTP layout is implemented in the BSM2-Python module bsm2_custom_layout.py, along with a customized initialization file bsm2_custom_layoutinit.py, a customized influent data file dyncustominfluent_bsm2.csv and a file to run the simulation custom_simulation.py. The implementation only uses BSM2-Python modules to simulate the WWTP layout. Therefore it is not possible to implement a perfectly identical model of the WWTP (without modelling of: screens, grits and grease chambers). Sludge inflows from outside into the WWTP were also not taken into account (but are still possible to implement). The biological stage is modelled with four activated sludge reactors in series with a single secondary clarifier/settler to simplify the layout and reduce the simulation time. For the simulation of the influent (flow rate (Q) [m³/d], total suspended solids (TSS) [g/m³], chemical oxygen demand (COD) [g(COD)/m³], biological oxygen demand after 5 days (BOD₅) [g(O₂)/m³], total Kjeldahl nitrogen (TKN) [g(N)/m³]) the BSM2-Influent-Generator was used.

We thank Nadja Hvala for her contribution by providing us with data and insights for the new implementation of the WWTP.