Kristopher Overholt

New web calculator – t-squared fire ramp generator

koverholt — Wed, 04 Jan 2012 00:05:57 +0000

Wondering when an ultrafast t-squared fire reaches 2,000 kW? 104 seconds! Or just want to quickly generate a slow t-squared curve to use in FDS? I’ve posted a new web calculator tool to generate t-squared fires, output plots, CSV files, and FDS syntax.

The t-squared fire ramp calculator is available here:

http://www.koverholt.com/t-squared-fire-ramp-calculator/

Please let me know if you find any bugs, would like to give feedback on this tool, or have a request for another web calculator!

How to compile CFAST on Mac OS X and Linux

koverholt — Wed, 20 Jul 2011 04:57:52 +0000

The Consolidated Model of Fire and Smoke Transport (CFAST) model is a two-zone fire model developed by NIST. From the NIST website, you can download the Windows version of CFAST, which includes a graphical user interface frontend (CEdit) for the creation and execution of CFAST cases. However, there are no downloadable CFAST binaries or user interfaces for the Mac and Linux platforms; you must compile CFAST yourself.

This guide will help you compile the CFAST executable on Mac OS X and Linux. This is useful for running CFAST cases on Mac or Linux machines, which is especially useful for scripting CFAST runs for optimization problems or running CFAST on a large number of cases in batch mode.

You will need the Intel Fortran compilers (ifort) and a copy of the CFAST source code from the CFAST Google Code site. You can try to use other free compilers, but I find that the Intel compilers are the most compatible and produce the most optimized (fastest) binaries. Note: there are heavily discounted Intel compilers for students.

To download the CFAST source code, the easiest way is to install subversion on your Mac or Linux machine and issue the following command:

svn co http://cfast.googlecode.com/svn/trunk/ cfast

Once you have the CFAST source code on your machine, you can perform the following steps to compile the CFAST executable.

1. In the cfast/CFAST/ directory, edit the makefile_linux file and make the following changes:

Replace ‘radation’ with ‘radiation’; this is a typo

Add ‘cyl_conduct.o’, ‘datamodules.o’, and ‘ssHeaders.f’ to the obj_serial section

Add the following lines to the Object Dependencies section:

cyl_conduct.o : cyl_conduct.f

datamodules.o : datamodules.f90

ssHeaders.o : ssHeaders.f

Add ‘cyl_conduct.f’, ‘datamodules.f90′, and ‘ssHeaders.f’ to the cfast.o line

Note: The corrected makefile can also be downloaded from here.

2. Copy datamodules.f90 from the cfast/CFAST/Include/ directory to the cfast/CFAST/Source/ directory

3. Change to the cfast/CFAST/Source/ directory

4. Run the command ‘ifort -c datamodules.f90′ to compile the datamodules and iofiles modules

5. Finally, run the command ‘make -f ../makefile_linux intel_osx_64’ to build CFAST

5b. If you are running Linux, the command is ‘make -f ../makefile_linux intel_linux’

Using these steps, I was able to build CFAST on an Intel Macbook Pro running 10.6 (Snow Leopard) as well as a Linux machine running CentOS release 5.6 with the 64-bit Intel Fortran Compiler version 12.0.4.184.

With your new CFAST executable file (cfast_osx_intel_64), you can run a CFAST case using the following syntax:

./cfast_osx_intel_64 casename

where casename is the name of your CFAST casename.in file (notice the .in file extension is omitted from the command).

Happy modeling!

Grassland fire research at The University of Texas at Austin

koverholt — Mon, 11 Apr 2011 06:54:03 +0000

While a large number of wildland fires are burning through West Texas and threatening the safety of life and property of Texans, the fire research group at The University of Texas at Austin is actively working with physics-based computer models and laboratory-scale fire tests to characterize the fuel properties and flame spread characteristics of grassland fuels.

Computer model of grassland fire simulation

At UT Austin, we have been performing small-scale, controlled experiments in our burn structure to determine ignition times and burning rates for grassland fuels as well as intermediate-scale, controlled experiments to determine the fuel and combustion properties of grass fuels, the effects of external wind on ember production, and the heat release rates of grass bunches.

Controlled grass fire test in the UT Austin burn structure

We can utilize the results of the small- and intermediate-scale experiments in full-scale computer simulations of grassland fires using modeling tools such as Wildland-urban interface Fire Dynamics Simulator (WFDS).

Using the results and methods from these controlled experiments along with the help of the wildland fire community, we plan to develop a framework to determine fuel properties of wildland fuels, predict the physics and fire dynamics behavior of wildland fires, and achieve safer conditions for people and property faced with the threat of wildland fire situations.

Inverse fire modeling for heat release rate characterization

koverholt — Wed, 23 Mar 2011 16:51:48 +0000

Attached are my PDF slides on the topic of “Inverse fire modeling for heat release rate characterization“, which was presented at the 7th US National Combustion Meeting in Atlanta, GA on March 21, 2011.

The abstract is as follows:

A ubiquitous source of uncertainty in fire modeling is the proper heat release rate for the fuel packages of interest. An inverse heat release rate (HRR) calculation method is presented to determine a HRR that satisfies measured temperature data. The methodology is developed by using synthetic temperature data using the Consolidated Model of Fire and Smoke Transport (CFAST) zone model to produce hot gas layer temperatures in a single compartment. The inverse HRR method runs at super-real-time speeds while calculating an inverse HRR solution that can reasonably well match the original HRR curve. Examples of the inverse HRR method are demonstrated by using a multiple step HRR case, experimental data with a constant HRR, and complex HRR curves. In principle, the methodology can be applied using any reasonably accurate fire model to invert for the HRR.

The slides can be downloaded here: Overholt_Combustion_2011

New web calculator – flame heights and plume centerline temperatures

koverholt — Thu, 10 Feb 2011 08:32:05 +0000

I’ve posted a new web calculator tool to calculate flame heights and plume centerline temperatures (above the flame height). The calculator is based on the correlations by Heskestad and McCaffrey, and is available here:

http://www.koverholt.com/flame-height-and-plume-centerline-temperature-calculator/

Please let me know if you find any bugs, would like to give feedback on this tool, or have a request for another web calculator!

New web calculator tool – transient steel heating under fire conditions

koverholt — Wed, 09 Feb 2011 04:55:29 +0000

I’ve posted a new web calculator tool related to structural fire safety to calculate the lumped temperature of steel under fire conditions. The tool has the ability to input parameters for steel, select a standard fire time-temperature curve (ISO or ASTM), and choose unprotected or protected steel. The tool is based on the equations from the textbook by A. Buchanan, Structural Design For Fire Safety, and is available here:

http://www.koverholt.com/transient-steel-heating-under-fire-conditions/

I hope you find the calculator useful, and I will continue to add web tools for fire protection engineering calculations. I will be adding a more detailed explanation and equations to the steel heating calculator page in the near future as well as releasing the source code.

Please let me know if you find any bugs, would like to give feedback on this tool, or have a request for another web calculator! The next web tool will be a calculator to give results from the Heskestad flame height and centerline temperatures.

Python ImportError with backend_wx2.8 and wxPython when using Matplotlib

koverholt — Tue, 11 May 2010 04:42:18 +0000

For many months, since I upgraded to Snow Leopard (10.6), I’ve had the following error when trying to import matplotlib. When I issued the command “import pylab” or “from pylab import *” in iPython, Python, or any other tool that uses the Python shell, I got the following error:

ImportError: Matplotlib backend_wx and backend_wxagg require wxPython >=2.8

Well, after poking around this problem for the thousandth time, I finally fixed it for good. I did this by removing my ~/.matplotlib/matplotlibrc file. In this file, there was only one line:

backend : WXAgg

which I removed (and since it was the only line, I just removed the file). Now the error is gone, and matplotlib imports just fine, meaning that I can use it in any Python shell, including Sage.

Just thought I’d post this up in case any other Python/Snow Leopard users are having trouble.

Flammable antifreeze in fire sprinklers allegedly causes explosion

koverholt — Thu, 25 Feb 2010 04:02:03 +0000

In August of 2009 in Truckee, CA, a man was cooking on a stove when the pan caught fire, activating the fire sprinklers. Upon activation of the sprinklers, a violent explosion occurred which resulted in 1 death and 4 burn injuries to the family.

I came across this copy of a report (from the IAFSS mailing list) which details a report of a cooking fire turned explosive when the sprinklers were activated. Allegedly the propylene glycol used in the fire sprinkler system as an antifreeze aided in the explosive event, and another case is cited in which the antifreeze ignited and assisted in growing the fire from its ignition source.

The report can be found here: California –explosion– report (PDF)

While I have taken a few courses in fire sprinkler systems, I am not extremely familiar with the flammability of antifreeze such as propylene glycol, but allegedly the atomization of the droplets causing the liquid to become flammable was the cause of this explosion. More flammability and MSDS hazard info is cited in the report.

The report presents many questions at the end which would be useful in quantifying the relationship of the antifreeze in this case, especially since (to my understanding) propylene glycol is the most widely used antifreeze in sprinkler systems. I look forward to finding out more about this topic, and I think it’d be a great research project for a student or research firm to take on.

The human element in fire protection engineering

M.S. thesis paper published

koverholt — Wed, 24 Feb 2010 18:01:14 +0000

My M.S. in Fire Protection Engineering from WPI was recently confirmed and my M.S. thesis was released for publication. The title of the thesis is “Characterizing the Flammability of Storage Commodities Using an Experimentally Determined B-number”.

Link to thesis is here: http://www.wpi.edu/Pubs/ETD/Available/etd-121409-192436/

The abstract is as follows:

In warehouse storage applications, it is important to classify the burning behavior of commodities and rank them according to material flammability for early fire detection and suppression operations. In this study, the large-scale effects of warehouse fires are decoupled into separate processes of heat and mass transfer. As a first step, two nondimensional parameters are shown to govern the physical phenomena at the large-scale, a mass transfer number, and the soot yield of the fuel which controls the radiation observed in the large-scale. In this study, a methodology is developed to obtain a mass-transfer parameter using mass-loss (burning rate) measurements from bench-scale tests. Two fuels are considered, corrugated cardboard and polystyrene. Corrugated cardboard provides a source of flaming combustion in a warehouse and is usually the first item to ignite and sustain flame spread. Polystyrene is typically used as the most hazardous product in large-scale fire testing. A mixed fuel sample (corrugated cardboard backed by polystyrene) was also tested to assess the feasibility of ranking mixed commodities using the bench-scale test method. The nondimensional mass transfer number was then used to model upward flame propagation on 20-30 foot stacks of Class III commodity consisting of paper cups packed in corrugated cardboard boxes on rack-storage. Good agreement was observed between the model and large-scale experiments during the initial stages of fire growth.

Characterizing Flammability of Corrugated Cardboard Using a Cone Calorimeter

Plotting data on videos – A useful way to convey qualitative and quantitive information

koverholt — Wed, 25 Nov 2009 08:49:44 +0000

Every once in a while, someone makes an impression on you that lasts for a lifetime. It sticks with you every single time. This is one of those, although a bit on the nerdy side, it is one that can change the way you present information in a very meaningful way.

I was once sitting at the NIST annual fire conference, going about my business, and someone working on a project regarding the structural response aspect of buildings on fire showed a video in their presentation. No big deal, right? Normally, we get cool fire videos, then some plots, and so on. Sometimes the plots are interesting, sometimes they are default from Excel with the ugly legend and all – with no story to tell.

But not this guy. He showed a video with real-time plots superimposed over the video showing the exact real-time structural response of the structure overlaid on the video in a plot. “AMAZING!” I thought. And it stuck with me. A useful way to convey synchronous information. People love videos, why not tell the qualitative AND quantitative story at the same time?

So I started working in grad. school on fire problems, and naturally, soon thereafter, I was scheduled to give a presentation. As most of my real creative coding and writing work happens of hours between the hours of 1 AM and 6 AM, I wanted to make this happen. I REALLY wanted some real-time plotting action in my presentation. No Excel templates for me! So I stayed up for a couple nights and worked on a way to use MATLAB to make this plotting dream a reality: I worked on importing videos, messing with frame rates, tons of images, and so forth. And soon thereafter, it happened. I had a working script.

I used it to show plots of large-scale fire tests with actual and predicted flame heights vs. time as seen here:

And I used the script to show the predicted flame heights on a small-scale test in an amazing way that just about anyone can relate to, fire-crazed scientist or not:

From anyone who has seen the videos firsthand, the response has been amazing. This is a great teaching and communication tool, and surprisingly enough, I haven’t found any existing program or tool that does this. And so I am sharing the videos and script here for anyone to use to better convey information.

My next steps are: 1) to convert the script to Python (since I am now almost exclusively using Python+numpy+scipy for my graduate research and daily work instead of MATLAB, and 2) to make the script into a cross-platform and easy to use tool.

I’m providing the code in its raw and uncommented and unedited form. It generates a number of images with plots superimposed on them, and then it is trivial to use a program to stitch them together into a video. I used Quicktime’s built in method. Sorry, too much current work going on finishing my MS thesis and Master’s degree to clean up the code, but it’s a brutal use of the “release early, release often” ideal! Hopefully someone can make some use of it.

So, here are the linked .m files:

http://www.koverholt.com/scripts/ssPlotVideo.m
http://www.koverholt.com/scripts/fireplotVideo.m

Enjoy! And please leave your comments or ideas!