Otwin

Otwin is an open-source physics engine designed for engineering systems, enabling the creation of digital twins from real equipment. It facilitates the integration of models built using Modelica, Julia, or components native to Otwin, allowing for a deeper analysis of machine behavior through empirical data.

Key Features

• Model Reliability Assessment: Otwin verifies the trustworthiness of engineering models by evaluating how accurate they are with respect to unseen data, identifying their predictive capabilities, and highlighting when actions should be taken based on model forecasts.

• Compatibility with Various Modeling Languages: Seamlessly integrate models derived from Modelica or Julia, or create comprehensive models using Otwin’s components.

• Forecasting and Evaluation: Otwin provides tools to assess forecast accuracy, producing conformal intervals that provide insights into uncertainty and reliability.

Example Usage

To install the Otwin engine:

pip install "otwin[engine]"

Real-World Application Scenarios

1. Chlorine Storage Tank Monitoring: Otwin was applied to track chlorine levels in a storage tank with predictions provided by a Modelica model. The insights obtained from Otwin indicated that the model underestimated chlorine levels, which proved crucial for maintaining safety standards:

   • Modelica predicted a stable level of chlorine, but Otwin revealed a more accurate range that aligned better with actual measurements.

   import numpy as np
   from fmpy import extract, simulate_fmu
   
   # Implementation details go here...
   

2. Water Level Forecasting in a Storage Tank: Using a ModelingToolkit model, Otwin assessed water levels and compared them against trivial forecasts, revealing discrepancies that informed operational decisions:

   • Observations showed the model significantly mispredicted peak water levels, underscoring Otwin's ability to identify subpar forecasting methods.

   import numpy as np
   from juliacall import Main as jl
   
   # Implementation details go here...
   

Build Models from Components

Otwin provides a rich set of components across various physical domains, enabling users to build systems from scratch. This allows creation of dynamic simulations that adhere to physical laws such as energy conservation:

import otwin
from otwin.components.mechanical import Mass, Spring, Damper, ForceSource, Fixed

mass = Mass(1.0)
spring = Spring(20.0)
damper = Damper(0.3)
weight = ForceSource(1.0 * 9.81)
ceiling = Fixed()

system = mass >> spring >> ceiling
system.connect(mass.flange, damper.a, weight.flange)
system.connect(damper.b, ceiling.port)

model = otwin.compile(system)
run = model.simulate(t_span=(0, 20), dt=0.05)

Documentation and Further Information

The documentation provides extensive details on component usage, integration steps, and examples validating model performance against real-world data. The project aims to facilitate contributions and ongoing development through community engagement on GitHub.

Explore the complete repository otwin for hands-on examples and further insights into the functionalities offered by the Otwin physics engine.