Comparison of DynSim and SysCAD

From a perspective of a Process Control Engineer wanting to:

simulate a combined control scheme and process model (where the process model non-linearities and dynamics that the control scheme sees is most important)
test more advanced control schemes (e.g. decoupling, constraint control, dead-time compensation / Smith-Predictors)

Note that this comparison is specifically NOT related to SysCAD's usefulness for its primary purpose (Process Engineering). Information on SysCAD has been sourced from the official documentation wiki at http://help.syscad.net/index.php/Main_Page.

General

Summary

add Matlab/Simulink here as well?

Feature	DynSim2	SysCAD
speed of execution (and edit-run cycle-time)	~1000x faster than SysCAD (based on 30min control simulation time) - edit-run cycle-time: 10 seconds	very slow - edit-run cycle-time: 1 hour?
ease of programming
integration control (e.g. for stiff models)
trend browsing (matlab style)	Yes
process control features of models
... dynamics
... output response to control influenced inputs	All models designed to respond to all control input variables, except where it would incur significant simulation burden (e.g. there are no size partitions)	cyclone - OK (Plitt)
process control scheme configurability
... range of built in blocks	PID, integrator, Smith-Predictor, constraint, filters (1st-order, MA), dead-time (for internal models), FOPDT, FIR, replay (of disturbances)	PID controller only
... advanced blocks (autocorrelation, FFT, ...)	Yes, anything in SciPy	No
... block useability for control simulations (e.g. tracking, velocity-mode, ...)
... ease of writing customised control blocks
... ease of combining blocks into a large control scheme
scriptability of overall simulation (for repetitive runs over a range of parameters)	Uses the already familiar and standard Python (already used as the simulation language)	Yes, using PGM (non-standard scripting language)
# of "languages" to learn (for full customisability)	1. Python (for simulations, scripting, new blocks)	1. graphical model language 2. PGM (for "scripting") 3. C++ (for new blocks)
interfacing (replay, logging, database, OPC)		* SysCAD has an OPC client (for gathering real-time data) and an OPC server (for allowing access by other tools to the simulation).
real-time
discrete-event
licence distribution (e.g. free packaging and distribution to clients)	Free! Simulation can be packaged in a convenient ".exe" file.	Client needs to buy a licence (see below)
cost
maintenance	self-maintained - easy to use language - only one language to learn	contract ($)

Speed of Simulation

SysCAD is not a slow process simulator. The models are coded in C++ allowing for the fastest execution possible, for the model functionality provided. What is important for a control engineer is whether the overheads of additional process calculations (size partitions, pressures, temperatures etc.) burden the simulation so much that it becomes cumbersome for a Process Control engineer, who is generally interested in the dynamics and non-linearities rather than the depth of the process calculations.

Specific Models

Control

PID
- no tracking or manipulation of bounds as part of a control scheme???
- cannot use gain scheduling as part of control scheme (tuning parameters are not connectable)

Process

pipe - SysCAD doesn't include any pipe dynamics, such as the propagation of a density change along a long pipe. This is important for density control loops which would otherwise have zero process dynamics.
thickener - no representation of bed and density dynamics due to UF pumping????