# Signal-flow vs Conserved System Modeling

### Mike Jensen

Posted Jan 10, 2011

There are many ways to approach system modeling and simulation. Two of the most popular methodologies are signal-flow and conserved. Naturally, the method you choose depends on the information you want from your simulation and analysis.

Just like the name implies, in a signal-flow description design information flows unidirectionally from input to output. While there might be multiple feedback paths within a signal-flow model, the inputs and outputs are essentially isolated from one another except for the mathematical processing described by the model itself. The interest is in how the input signal gets processed to produce the output signal; how the system consumes energy isn’t considered. Signal-flow descriptions are great for modeling the time-based relationships between measurable entities (e.g. differentiating velocity to get acceleration), but lack the ability to accurately model a system’s physics-based peformance.

Contrast signal-flow with conserved system modeling. Conserved modeling deals with device physics and accounts for how energy is consumed in a system; the energy supplied to, and consumed by, a system must sum to zero. As an example, the performance of electrical circuits must obey two important rules: Kirchoff’s Voltage Law (KVL) and Kirchoff’s Current Law (KCL). KVL simply states that voltages around a closed loop must sum to zero: a battery or similar voltage source adds voltage to a circuit, and any type of load subtracts voltage from a circuit. KCL states a similar rule for currents in an electrical circuit: the algebraic sum of currents entering and leaving a circuit node must be zero. A key benefit of conserved modeling is the ability to analyze system loading effects — how one block loads another — information not available from a signal-flow analysis. Analogous conservation laws exist for other energy domains.

When assembling a system model, the modeling method used determines the information a simulation can give you about your system. Naturally, there are times when signal-flow modeling is useful, and times when conserved modeling makes the most sense.  And there are times when using both signal-flow and conserved blocks in a single system model yields the best and most complete analysis. The challenge is finding a modeling and simulation platform that supports standalone signal-flow and conserved system descriptions, as well as an option to combine both methodologoes in single system desription. SystemVision coupled with the IEEE standard VHDL-AMS modeling language is among the handful of industry available solutions for standalone, and integrated, signal-flow and conserved system modeling.

In a future blog post, I’ll describe a modeling technique that uses the power of signal-flow and conserved modeling to describe system functions — all without writing a single line of model code.

### More Blog Posts

The challenge is finding a modeling and simulation platform that supports standalone signal-flow and conserved system descriptions, as well as an option to combine both methodologoes in single system desription. Finding such simulator is not that difficult: already available for many years, and since 2010 also available as OSCI standard: SystemC AMS 1.0 http://www.systemc.org/downloads/standards/ams10 Supports data-flow, signal-flow and conservative networks in a single-kernel fashion. Regards, Martin

Martin
9:01 PM Jan 10, 2011

You are correct, Martin, in pointing out there are several simulation platforms (simulator + modeling language) available that claim support for signal-flow and conserved modeling and simulation. Some have been around for a number of years (note that developing and fine-tuning such a simulation platform is not an overnight project, so most viable options in this class have, by definition, been around for awhile). Most simulators/languages are targeted at a specific solution space, although the combination may solve general problems within the target technology, and can sometimes be effectively applied to problems in other technologies. SystemC is largely targeted at embedded system development, though the SystemC AMS 1.0 standard is a nice addition that extends it's capabilities. Though I didn't specifically mention it in my blog post, my comments were more focused on mechatronic system modeling and simulation -- which naturally might include SystemC descriptions of certain blocks. There are just a handful of viable platform options (SystemVision, Saber, Simplorer, etc.) for efficient mechatronic system modeling, simulation, and analysis. Thanks for the comment!

Mike Jensen
7:53 PM Jan 12, 2011