pyBondGraph

pyBondGraph is a Python library for modeling and analyzing linear bond graph systems using symbolic computation.

The library allows users to construct bond graph models programmatically, automatically derive the governing equations, and analyze the resulting dynamic systems using tools from control theory.

Bond graphs provide a domain-independent modeling framework for physical systems. Using a unified representation of power exchange, the same modeling approach can be used for electrical, mechanical, hydraulic, and multi-domain systems.

---

Features

• Programmatic construction of bond graph models
• Automatic causality assignment via SCAP (Sequential Causality Assignment Procedure), with optional manual override or mixed mode
• Automatic symbolic equation derivation using SymPy
• Conversion of models to linear state-space systems ($\dot{x} = Ax + Bu$, $y = Cx + Du$)
• Composable sub-models via SubBondGraph with deep-copy namespace isolation
• Two-port elements: Transformer and Gyrator with automatic causality propagation
• Sensor elements: IntegratedEffortSensor and IntegratedFlowSensor for measuring integrated generalized variables (e.g. position from velocity)
• Domain-neutral aliases: Compliance = Capacitor, Inertance = Inductor, Resistance = Resistor
• Example models for electrical and electromechanical systems
• Integration with python-control for numerical simulation (step response, Bode plots, etc.)

---

Installation

Install from PyPI or Github

The easiest way to install the library is via your preferred package manager (e.g. pip) directly from PyPI:

pip install pyBondGraph

Alternatively one can install the latest development version directly from the GitHub repository:

pip install git+https://github.com/MrP123/pyBondGraph.git

---

Development installation

To work with the source code:

git clone https://github.com/MrP123/pyBondGraph.git
cd pyBondGraph
pip install -e .

---

Dependencies

The main dependencies are:

• sympy
• numpy
• networkx
• matplotlib
• control only needed for the examples

Optional dependencies are used for experimental visualization tools.

---

Basic Usage

A bond graph model is constructed by creating elements and connecting them via the connect() convenience method, which creates bonds and adds them to the graph in one step.

RC-Filter with automatic causality (SCAP)

from pyBondGraph import BondGraph, SourceEffort, Resistor, Capacitor, OneJunction

bg = BondGraph()

# create elements
voltage_source = SourceEffort("U", "u_in")
resistor = Resistor("R", "R")
capacitor = Capacitor("C", "C")
series_junction = OneJunction("J1")

# connect elements --> causality is assigned automatically by SCAP
bg.connect(voltage_source, series_junction)
bg.connect(series_junction, resistor)
bg.connect(series_junction, capacitor)

# plot the resulting BondGraph
bg.plot()

# derive system equations in linear state space form
A, B, C, D, x, n_states, n_inputs, n_outputs = bg.get_state_space()

Causality can also be assigned manually by passing a Causality value to connect(), or in mixed mode where some bonds are fixed and SCAP resolves the rest.

The library automatically derives the symbolic system equations describing the dynamics of the model.

---

Core Concepts

Bond graphs represent power exchange between system components, where power is the product of effort and flow associated with the following components:

Elements

Element	Meaning
R	Dissipation
C	Energy storage (compliance, capacitance)
I	Energy storage (inertia, inductance)
Se	Effort source
Sf	Flow source

Two-Port Elements

Element	Meaning
TF	Transformer — same causality on both bonds
GY	Gyrator — opposite causality on both bonds

Junctions

Junction	Meaning
0	Common effort
1	Common flow

Sensors

Sensor	Meaning
IntegratedEffortSensor	Measures integral of the effort at its bond
IntegratedFlowSensor	Measures integral of the flow at its bond

In mechanical bond graph models:

• flow corresponds to velocity, i.e. an IntegratedFlowSensor can be used to compute position.
• effort corresponds to force

---

Example Systems

The repository contains example models illustrating typical applications of bond graphs.

RLC Circuit

Demonstrates modeling of an electrical circuit using bond graph elements.

DC Motor

A multi-domain electromechanical system coupling electrical and mechanical dynamics. Also demonstrates integration with the python-control package for numerical simulation (step response).

Transformer

Example of energy transformation between two ports.

Two DOF Mass–Spring–Damper System

Classical mass-spring-damper system with two degrees of freedom.

---

Causality Assignment

pyBondGraph supports three modes for assigning causality:

1. Automatic (SCAP) — omit causality in connect() calls; assign_causality() is called automatically when solving. The Sequential Causality Assignment Procedure assigns causality in priority order: sources, storage elements (integral causality), resistors, then propagation through junctions and two-port elements.
2. Manual — pass Causality.EFFORT_OUT or Causality.FLOW_OUT explicitly to each connect() call.
3. Mixed — fix causality on some bonds, let SCAP resolve the rest.

If a storage element cannot receive integral causality (which would imply a DAE rather than an ODE), a DerivativeCausalityError is raised with a clear diagnostic message.

---

Typical Applications

Bond graph modeling is particularly useful for:

• electromechanical systems
• robotics and mechatronics
• multi-domain energy systems
• control system modeling
• teaching system dynamics

---

Planned Features

• FMU Export: export bond graph models as Functional Mock-up Units (FMI standard) for interoperability with Simulink, Dymola, OpenModelica, and other FMI-compliant tools
• Nonlinear element support: general nonlinear constitutive laws with Jacobian linearization
• Convenience bridge to python-control: to_control_ss(params) method wrapping the existing manual pattern