DWSIM Dynamic Simulation with Python

Automating process dynamic simulations from Python using the DWSIM Automation API and pythonnet.

This repository contains a series of case studies showing how to drive DWSIM dynamic flowsheets programmatically. Each case is self-contained and builds on the same core idea: use Python to make dynamic process simulation reproducible, scriptable, and easier to connect with larger computational workflows.

---

Why scripted dynamic simulation?

Running a dynamic simulation from the DWSIM GUI is useful for interactive exploration. It becomes limiting when the goal is to repeat the same scenario many times, run structured what-if studies, generate datasets, or compare control strategies.

Once a dynamic flowsheet can be controlled from Python, it becomes a reproducible process scenario. The same model can be used to run parameter sweeps, test controller behaviour, generate time-series data, or evaluate how the system responds to structured disturbances — all without manually operating the GUI.

---

Repository structure

dwsim-python-dynamic-simulation/
├── README.md
├── 01_separator/
│   ├── dwsim_separator_dynamic_control.ipynb
│   └── separator_python_run.csv
└── 02_separator_from_scratch/
    ├── DWSIM_SEPARATOR_REFERENCE_MODEL.ipynb
    ├── DWSIM_SEPARATOR_PYTHON_BUILT_MODEL.ipynb
    └── results/

---

01 · Gas-Liquid Separator — Automating an existing simulation

The first case study automates an existing DWSIM dynamic sample flowsheet from Python.

Instead of running the dynamic schedule from the DWSIM GUI, the notebook loads the model, advances the dynamic solver step by step, applies pressure setpoint changes manually, triggers the internal PID controllers, and records the response.

Base model: Dynamic Simulation - Separator Pressure and Level Control.dwxmz Property package: Raoult's Law Compounds: Air, CO₂, Water, Methanol

Flowsheet

Feed ──► FV-01 ──► SG-01 ──► PV-01 ──► gas outlet
                      │
                      └──► LV-01 ──► liquid outlet

Control structure

Controller	Controlled variable	Manipulated variable	Setpoint
PID-012	Separator pressure	PV-01 opening	2 bar
PID-013	Separator liquid level	LV-01 opening	0.3 m

Scenario

A 30-minute dynamic run with two scheduled pressure setpoint changes:

• t = 3 min → pressure setpoint steps to 5 bar
• t = 10 min → pressure setpoint returns to 2 bar

Contents

01_separator/
├── dwsim_separator_dynamic_control.ipynb   ← main notebook
└── separator_python_run.csv                ← recorded time series

Result

The 30-minute dynamic scenario runs in approximately 50 seconds, around 36× faster than real time.

Key findings

Topic	Correct pattern	Common mistake
Dynamic solver	DWSIMSolver.SolveFlowsheet(fs, 1)	RequestCalculationAndWait runs the steady-state solver
Schedule execution	Reproduce events manually in Python	dm.RunSchedule() is GUI-dependent
Event trigger time	ev.TimeStamp	ev.TriggerTime does not exist
Event object reference	ev.SimulationObjectID is a GUID and needs lookup	Using it as a tag directly
PID setpoint	pid.SetPoint	pid.SetPointAbs does not exist
Monitored variable history	Capture values inside the Python loop	Expecting history to persist after reload

---

02 · Gas-Liquid Separator — Building a dynamic model from scratch

The second case study goes one step further: instead of only automating an existing flowsheet, it builds an equivalent DWSIM dynamic separator model programmatically from Python.

A reference model is first prepared from the DWSIM built-in sample case. Then, a second notebook constructs the equivalent flowsheet from scratch and validates it against the reference behavior.

Property package: Raoult's Law Compounds: Air, CO₂, Water, Methanol Design basis: 7,500 kg/h Vessel: 1.0 m³ volume, 2.0 m height Main setpoints: pressure = 2.0 bar, liquid level = 1.0 m

Valve sizing

Valve	Fluid/service	Kv	Basis
PV-01	Gas outlet	85	IEC 60534-2-1 gas sizing
LV-01	Liquid outlet	12	IEC 60534 liquid sizing
FV-01	Feed inlet	450	Flow boundary / inlet pressure drop control

Notebooks

DWSIM_SEPARATOR_REFERENCE_MODEL.ipynb

Starts from the DWSIM built-in separator dynamic sample and prepares a reference case for validation.

The workflow:

• removes the internal DWSIM PID controllers and display instruments;
• resizes the valves using IEC-based Kv values;
• fixes outlet boundary pressures;
• pre-equilibrates the model using Python PID controllers;
• saves the prepared reference state;
• runs open-loop and closed-loop dynamic tests.

Dynamic tests include:

• open-loop feed step;
• closed-loop pressure setpoint changes;
• closed-loop level setpoint changes.

DWSIM_SEPARATOR_PYTHON_BUILT_MODEL.ipynb

Builds an equivalent dynamic separator flowsheet directly from Python, without manual GUI construction.

The workflow:

• creates the material streams and unit operations;
• defines the thermodynamic package and compounds;
• connects the flowsheet topology;
• configures valve sizing and separator geometry;
• initializes the dynamic state;
• runs the same scenarios as the reference model;
• compares the Python-built model against the reference model.

Contents

02_separator_from_scratch/
├── DWSIM_SEPARATOR_REFERENCE_MODEL.ipynb       ← reference model preparation and tests
├── DWSIM_SEPARATOR_PYTHON_BUILT_MODEL.ipynb    ← Python-built model and validation
└── results/                                    ← generated locally, not versioned

The results/ folder is kept in the repository with a .gitkeep file, but generated plots and temporary outputs are not committed. They can be regenerated by running the notebooks.

Key findings

Finding	Detail
Python-built valves require explicit Kv/opening behaviour	The opening-Kv relationship may need to be enabled explicitly through reflection.
Dynamic properties must be refreshed after each solver step	sg.RunDynamicModel() is required after each step to read updated pressure and level values reliably.
Saved flowsheets may contain orphaned graphic objects	Removed GUI objects can leave XML nodes behind unless the saved file is cleaned.
Initial valve opening matters	Starting too close to the hydraulic limit can destabilize the pre-equilibration.
Vessel pressure initialization matters	The inlet stream should be initialized near the operating pressure before the first dynamic step.
PI control is preferable for setpoint tracking	Derivative action can create spikes during setpoint changes.

---

Requirements

• DWSIM 9.x installed on Windows
• Python 3.x
• pythonnet 3.x
• numpy
• pandas
• matplotlib

Install Python dependencies:

pip install pythonnet numpy pandas matplotlib

---

Related

• DWSIM — open-source process simulator
• LinkedIn article series — process engineering context and discussion for each case study

---

DWSIM 9.x · Python 3.x · pythonnet 3.x