Upcoming plans for CFAST

[mcgratta]

I am writing this post to solicit feedback from the CFAST user community about upcoming plans to overhaul several components of the software. These changes are necessary for us to maintain the software now that its original developers have retired. They still help out with bugs and other issues, but it is clear that some modernization is necessary to maintain the software's viability.

First of all, we have no plans at the moment to change the core CFAST solver; that is, the Fortran program itself. Rather, we plan the following:

1. We are currently converting the graphical user interface (GUI) from Visual Basic (VB) to Python. This will allow the GUI to run on all platforms, not just Windows, and we envisage that a program like CFAST would be ideal for tablets, phones, and whatever else comes along in the coming years. With VB, we had few pathways for growth, and, worse, Cedit, the VB interface, relies on a piece of proprietary software for certain features that are now routine with various Python modules.
2. We have converted the verification and validation (V&V) scripts from Matlab to Python, and will continue to convert the various scripts used for packaging and distributing new releases.
3. We would like to know, and this Discussion thread can serve as a convenient forum, if the companion program called Cdata is widely used. This program is essentially a wrapper that enables a user to launch a collection of CFAST cases where various input parameters are varied. We have heard anecdotally that some CFAST users have developed their own means of doing this, and we want to know if Cdata is worth maintaining. At the moment, it is not being actively supported.
4. We have no immediate plans to replace Smokeview as the visualization tool for CFAST, although we'd be interested if any of you have experimented with something like this.

To sum up, we know that CFAST is still a useful tool and we are committed to maintaining it. However, because CFAST is not under active development, we do not receive as much feedback on its use compared to FDS, for which we hear from users everyday about various issues. These proposed changes will enable us to maintain the software with less effort and make it easier to use.

[boris-stock]

Kevin,

just my two cents:

3. I have to confess I haven't used Cdata yet, but it is right on top of my to-do list. I think it is highly interesting, the next important step in documentation of zonemodells. I heard of no specific implementation, which can be publicly used.
4. I know of none alternative to Smokeview...

Boris

[bwklein]

We have discussed adding CFAST solver nto one of our GUI tools. Ventus a CONTAM GUI is probably the closest conceptually to CFAST, as you define zones(room volumes), properties of them, and the connections between them. With a few more UI elements I could see this tool being able to build CFAST input files, run the simulator, and then view the results in the same UI. Just like we do now for CONTAM.

I'm not sure if we would need to integrate the Cdata tool, as we would probably have our own method for spawning multiple simulations from various scenarios.

Jason F. mentioned to me a while ago that JH had developed their own zone model for internal use, that it had features missing from CFAST. It might be interesting to see what those are and add them into CFAST solver, making it more useful for a larger user base.

[mcgratta]

Our plans are simply to modernize the GUI and packaging. We have no plans to change CFAST itself. The new GUI would work like the old one, but it would be something we can fix if need be.

[bewygs]

Hi Kevin,

We use CFAST extensively at our company, and over time we've built lots of workflows around it.

Regarding CData, we did try to use it but ran into several bugs related to inputs we wanted to vary, particularly the radiative fraction of the fire. We ended up building our own python script, which eventually became PyCFAST. So yes, the use case CData addresses is very important to us, but its limitations led us to build our own tool.

Smokeview is really good, no need to change it.