Fixed Step vs Variable Step Solver in Simulink: Key Differences and Usage

In Simulink, a fixed step solver uses a constant time step for simulation, providing predictable timing but less flexibility. A variable step solver adjusts the time step dynamically for better accuracy and efficiency, especially in complex or stiff models.

Quick Comparison

This table summarizes the main differences between fixed step and variable step solvers in Simulink.

Factor	Fixed Step Solver	Variable Step Solver
Time Step	Constant throughout simulation	Changes dynamically based on model behavior
Accuracy	Lower, depends on fixed step size	Higher, adapts to system dynamics
Simulation Speed	Faster for simple or real-time systems	Potentially slower but more efficient for complex models
Use Case	Real-time systems, hardware-in-the-loop	Non-real-time, stiff or complex systems
Determinism	Highly deterministic timing	Less deterministic due to variable steps
Memory Usage	Predictable memory use	May use more memory due to adaptive steps

Key Differences

The fixed step solver advances the simulation by a constant time interval, which makes it ideal for real-time applications where timing predictability is crucial. However, this can lead to less accurate results if the step size is not small enough to capture fast system changes.

On the other hand, the variable step solver changes the step size during simulation based on the system's dynamics. It uses smaller steps when the system changes rapidly and larger steps when changes are slow, improving accuracy and efficiency. This adaptability makes it suitable for complex or stiff systems but less ideal for real-time constraints.

Choosing between them depends on the trade-off between simulation speed, accuracy, and real-time requirements. Fixed step solvers guarantee consistent timing, while variable step solvers optimize step size for better precision.

Code Comparison

Here is an example of setting up a fixed step solver in Simulink using MATLAB commands.

matlab

set_param('model_name', 'Solver', 'ode3');
set_param('model_name', 'FixedStep', '0.01');
set_param('model_name', 'SolverType', 'Fixed-step');

Output

Simulink model 'model_name' configured to use fixed step solver with step size 0.01 seconds.

Variable Step Solver Equivalent

Here is how to configure the same Simulink model to use a variable step solver.

matlab

set_param('model_name', 'Solver', 'ode45');
set_param('model_name', 'SolverType', 'Variable-step');

Output

Simulink model 'model_name' configured to use variable step solver ode45.

When to Use Which

Choose a fixed step solver when you need consistent timing, such as in real-time simulations, hardware-in-the-loop testing, or when the model runs on embedded systems. It is simpler and ensures predictable execution intervals.

Choose a variable step solver when accuracy is more important than timing consistency, especially for complex, nonlinear, or stiff systems. It adapts step size to system behavior, improving simulation efficiency and precision.

Key Takeaways

Fixed step solvers use constant time steps, ideal for real-time and predictable timing needs.

Variable step solvers adjust step size dynamically for better accuracy and efficiency.

Use fixed step for hardware-in-the-loop and embedded simulations.

Use variable step for complex or stiff models requiring precise results.

Choosing the solver depends on the trade-off between timing determinism and simulation accuracy.