Skip to content

MaCFP 2023 Modeling Guidelines

Randy McDermott edited this page Aug 30, 2023 · 68 revisions

Guidelines for Participation in the Third Workshop Organized by the MaCFP Working Group – 22 October 2023, Tsukuba, Japan

Submissions of computational results are due August 31

Please go to this link for instructions on Submitting Computational Results.

Submissions of posters are due October 8

In coordination with submission of computation results, please email your poster to the respective session co-chairs and technical points of contact (see below) with cc to Arnaud, Bart, Isaac, and Randy. Please include a PDF and the raw images (e.g., via Power Point) so the co-chairs may pull from your poster do develop their discussion talks. Bring your poster to the MaCFP-3 workshop and plan to arrive early to set up your poster prior to the meeting.

Introduction

The general objective of the “IAFSS Working Group on Measurement and Computation of Fire Phenomena” (abbreviated as the “MaCFP Working Group”) is to establish a structured effort in the fire research community to make significant and systematic progress in fire modeling, based on a fundamental understanding of fire phenomena. This is to be achieved as a joint effort between experimentalists and modelers, identifying key research topics of interest as well as knowledge gaps, and thereby establishing a common framework for fire modeling research. The MaCFP Working Group is intended as an open, community-wide, international collaboration between fire scientists. It is also intended to be a regular series of workshops. The first MaCFP workshop was held on June 10-11, 2017, as a pre-event to the 12th IAFSS Symposium in Lund, Sweden. The second MaCFP workshop was held on April 22-23, 2021, as a pre-event to the 13th IAFSS Symposium. Details on the content and outcomes of the first and second MaCFP workshops (called MaCFP- 1 and MaCFP-2) can be found on the MaCFP website and MaCFP repository.

Third MaCFP Workshop ("MaCFP-3")

MaCFP-3 will take place on October 22, 2023, as an (in-person) pre-event to the 14th IAFSS Symposium in Tsukuba, Japan. The workshop will feature activities organized by the Gas Phase Phenomena subgroup, the Condensed Phase Phenomena subgroup, as well as the recently created Radiative Heat Transfer Phenomena subgroup. We focus here on activities of the Gas Phase and Condensed Phase Phenomena subgroups.

As previously announced, the third MaCFP workshop will feature comparisons of experimental data and computational results obtained in separate, decoupled, condensed-phase and gas-phase fire configurations as well as in fully coupled cases corresponding to flame spread over a combustible solid. The Gas Phase and Condensed Phase Phenomena subgroups are planning to hold discussions corresponding to the following target experiments:

  • NIST-Waterloo-Pool-Fires: 30 cm, 37 cm, and 100 cm diameter liquid pool and gaseous burners studied at NIST and featuring multiple fuels;

  • FM-Burner: 13.7 cm diameter ethylene diffusion flames studied at FM Global and featuring a controlled co-flow oxygen-nitrogen oxidizer;

  • NIST-Gasification-Apparatus: bench-scale thermal degradation experiments conducted in the NIST gasification apparatus, providing validation data for PMMA pyrolysis models;

  • UMD-SBI: Flame spread experiments in a 1.46 m corner wall configuration studied at the University of Maryland with MaCFP PMMA (based on the Single Burning Item (SBI) Test, EN13823);

  • NIST-Parallel-Panel: Flame spread experiments in a 2.44 m parallel panel configuration studied at NIST with MaCFP PMMA (based on the FM4910 Parallel Panel Test).

The MaCFP repository hosted on GitHub contains all available experimental data corresponding to these different configurations. Note that the repository is continuously updated, and users are expected to consult the repository regularly on possible additions and/or corrections. These data are available to computational groups for model validation. For MaCFP-3, Python-based post-processing tools are available here. However, contributors need only to submit comma-delimited (*.csv) files with their computational results together with another comma-delimited configuration file listing the plots to be made (see Submitting Computation Results). An example is provided in the wiki. Please contact Randy McDermott if you have any questions regarding submission of results or post-processing.

MaCFP-3 will present detailed comparisons between experimental data and computational results obtained by participating modeling groups with the intent to review progress, summarize accomplishments, identify knowledge gaps, and provide guidance with clear objectives for the next workshop. The spirit in which discussions will be conducted is collaborative and collegial.

Interested modeling groups should inform the Gas Phase and Condensed Phase Phenomena subgroups of their plans to participate in MaCFP-3 by contacting the following Co-Chairs:

  • Morgan Bruns, [email protected] (Co-Chair of the Condensed Phase Phenomena subgroup) (NIST-Gasification-Apparatus, UMD-SBI, NIST-Parallel-Panel)

  • Isaac Leventon, [email protected] (Co-Chair of the Condensed Phase Phenomena subgroup) (NIST-Gasification-Apparatus, UMD-SBI, NIST-Parallel-Panel)

  • Bart Merci, [email protected] (Co-Chair of the Gas Phase Phenomena subgroup) (NIST-Waterloo-Pool-Fires, FM-Burner, UMD-SBI, NIST-Parallel-Panel)

  • Arnaud Trouvé, [email protected] (Co-Chair of the Gas Phase Phenomena subgroup) (NIST-Waterloo-Pool-Fires, FM-Burner, UMD-SBI, NIST-Parallel-Panel)

Approximately 6 weeks before the Workshop (i.e., by the end of August 2023), participating modeling groups will be asked to submit an electronic copy of their computational results organized in simple comma-delimited ASCII files. All results will be uploaded on the MaCFP GitHub repository and detailed comparisons between experimental data and computational results will be compiled by the MaCFP organizers in collaboration with experimentalists and modelers.

We present below guidelines for submission of computational data by participating modeling groups.

Timeline of Events in Preparation for MaCFP-3

All participants who wish to participate in MaCFP-3 by submitting modeling results are asked to prepare a poster summarizing their work for display at the Workshop. One-page poster abstracts should be electronically submitted by June 30, 2023: Submit MaCFP Poster Abstract.

  • June 1: Deadline to submit modeling results for the MaCFP PMMA pyrolysis modeling validation exercise. Details summarized below and on the [link: MaCFP Github Repository Release]
  • June 20: MaCFP-3 Virtual Seminar (Tuesday, 12:00 pm EST): “Presentation of pyrolysis models (i.e., material property sets) for use in the MaCFP-3 Workshop (UMD-SBI and NIST-Parallel-Panel Tests)”. Register here to receive a confirmation email message that contains the meeting password and instructions for joining the meeting
  • June 30: Poster abstract submission deadline (abstract of modeling submissions for the MaCFP-3 Workshop). Please note that poster submission is mandatory for participation in MaCFP-3. Posters will be available for discussion at the Sunday workshop and also during the regularly scheduled IAFSS poster session. When submitting your poster abstract, please look for and check the box indicating you are submitting a MaCFP poster. This will ensure space at the IAFSS poster session and help the poster chairs select appropriate reviewers.
  • August 31: Deadline to submit modeling results (pool fires, gaseous burners, fire growth cases). Late results may be included on final posters, but we cannot guarantee that these results will be directly included in the discussion talks.

General Guidelines

We ask that in MaCFP-3 submitted computational results for any gas phase or coupled simulations (i.e., NIST-Waterloo-Pool-Fires, FM-Burner, UMD-SBI, NIST-Parallel-Panel) comply with the following series of requirements:

  • We ask that for each simulated target experiment, the submission includes a grid convergence study in which the effect of changing spatial resolution in the flow and combustion solver (as opposed to the radiation solver, discussed next) is quantified;

  • Similarly, we ask that for certain simulated target experiments, the submission includes an angular convergence study in which the effect of changing angular resolution in the radiation solver is quantified;

  • We ask that modeling groups explain their modeling choices for the treatment of the turbulent flow, combustion and radiation transport; we encourage modeling groups to define a baseline model and apply that model to all simulated cases considered by the group; we ask that variations in modeling choices be justified.

We believe that these requirements will improve the quality and depth of the comparisons between results obtained by different modeling groups.

Furthermore, we encourage modeling groups to consider performing fine-grained simulations under the high-resolution conditions that are often the preferred choice made by CFD researchers, and also to consider performing coarse-grained simulations under the moderate-to- marginal resolution conditions (sometimes called VLES) that are more representative of the choices made by CFD practitioners.

NEW Please present the information above in a poster to be displayed at the workshop and submitted as a pdf document for archival on the GitHub site.

Additional Guidelines for NIST-Waterloo-Pool-Fires

NIST-Waterloo-Pool-Fires corresponds to 30 cm, 37 cm, and 100 cm diameter pool and gaseous burner flames studied at NIST (see documentation provided here) and the University of Waterloo (see documentation here) including:

  1. 30 cm Methanol Liquid Pool Fire [Waterloo and NIST]
  2. 30 cm Ethanol Liquid Pool Fire [NIST]
  3. 30 cm Acetone Liquid Pool Fire [NIST]
  4. 100 cm Methanol Liquid Pool Fire [NIST]
  5. 37 cm Methane Gaseous Pool Fire (34 kW) [NIST]
  6. 37 cm Propane Gaseous Pool Fire (20 kW) [NIST]
  7. 37 cm Propane Gaseous Pool Fire (34 kW) [NIST]

Evaporating liquid pool fires should be simulated with both prescribed and predicted mass loss rate.

Suggested grid resolution (30 cm pool, 37 cm burner): ∆x from 0.5 cm to 6 cm.

Suggested grid resolution (100 cm pool): ∆x from 1 cm to 20 cm.

Suggested angular resolution: NΩ = 100 angles.

Plots of interest include:

  • Plots showing the mean and rms centerline temperature as a function of elevation z (pool fires #1 - #5);
  • Plots showing the mean and rms vertical velocity as a function of elevation z (pool fires #1 - #5);
  • Plots showing the mean and rms temperature as a function of radial position r, for different elevations z (pool fires #1 - #3);
  • Plots showing the mean and rms vertical velocity as a function of radial position r, for different elevations z (pool fires #1 and #2);
  • Plots showing the mean and rms radial velocity as a function of radial position r, for different elevations z (pool fires #1 and #2);
  • Plots showing the mean centerline volume fractions of selected gas species as a function of elevation z (pool fires #1, #2, #4, and #5);
  • A plot showing the time variations of the heat release rate (pool fires #1 - #5);
  • Representative plots showing the instantaneous flame shape (e.g., identified as the 200- kW/m3 iso-contour of the volumetric heat release rate, or using some other method to be specified) (pool fires #1 - #5);
  • A plot quantifying the puffing frequency of the pool fire (pool fires #1 - #5);
  • Plots showing the radial variations of the total heat flux for gauges oriented horizontally, facing upwards, and located on the plane defined by the burner rim, within and beyond the pool burner (pool fires #1 - #3);
  • Plots showing the vertical variations of the radiative heat flux for gauges oriented vertically, facing the centerline of the fire, and located at r = 60 cm (pool fires #1, #2) and r = 207.5 cm (pool fire #3).

Additional Guidelines for FM-Burner

FM-Burner corresponds to a 13.7 cm diameter, controlled coflow (a mixture of air and nitrogen), round ethylene diffusion flame experiment currently studied at FM Global (see documentation here). FM-Burner includes global measurements at coflow oxygen mole fractions of 20.9%, 19%, 17% and 15%, and detailed measurements at coflow oxygen mole fraction of 20.9%, 16.8% and 15.2%. Groups interested in simulating FM-Burner are invited to predict the changes in flame radiation. Simulations with a prescribed radiant fraction (taken as the measured averaged value) are acceptable as well.

The experimental database includes combustion efficiency as a function of coflow oxygen mole fraction for four different fuels: methane, ethylene, propylene, and propane. Modelers are encouraged to predict the limiting oxygen concentration at extinction for each fuel and for a range of grid resolutions, as local extinction may have a significant effect on radiative emission, and these global extinction comparisons may provide insight about grid dependence of the combustion model.

Suggested grid resolution: ∆x from 0.5 cm to 2 cm.

Suggested angular resolution: NΩ = 100 angles.

Plots of interest include (for ethylene only):

  • Plots showing the mean and rms temperature as a function of radial position r, for different elevations z;

  • Plots showing the probability density function (PDF) of temperature for different radial positions r and elevations z;

  • Plots showing the mean and rms soot volume fraction as a function of radial position r, for different elevations z (for comparisons, use the experimental soot data corresponding to Laser Induced Incandescence – LII – measurements);

  • Plots showing the probability density function (PDF) of soot volume fraction for different radial positions r and elevations z (for comparisons, use the experimental soot data corre- sponding to Laser Induced Incandescence – LII – measurements);

  • A plot showing the time variations of the heat release rate;

  • Representative plots showing the instantaneous flame shape (e.g., identified as the 200 kW/m3 iso-contour of the volumetric heat release rate, or using some other method to be specified);

  • A plot showing the variations of the predicted global radiant fraction with the coflow oxygen mole fraction;

  • Plots showing the vertical variations of the radiative emission power per unit height of the fire (kW/m) (in numerical simulations, this quantity can be estimated by calculating the average of the radiation source term that appears in the energy equation integrated over each horizontal plane and over time).

Plots of interest include (for all fuels):

  • A plot of combustion efficiency as a function of coflow oxygen mole fraction

Additional Guidelines for NIST-Gasification-Apparatus

NIST-Gasification-Apparatus corresponds to bench-scale thermal degradation experiments of PMMA conducted at NIST (see validation data for PMMA here and for the NIST gasification apparatus here). NIST-Gasification-Apparatus includes measurements of sample mass and back surface temperature for two values of the radiant heating (25 kW/m2 and 50 kW/m2) and in an anaerobic environment (nitrogen). Additional tests performed with inert materials (copper and Kaowool PM insulation) are also included; these test results may be used to validate the modeling choices made for material thermophysical properties and boundary conditions (e.g., convection heat transfer).

Participants who submitted pyrolysis models (i.e., material property sets for MaCFP PMMA) to the MaCFP-2 Workshop are asked to use their original model parameters to predict (without adjustment to material properties) the results of the NIST-Gasification-Apparatus validation experiments. These blind predictions should be shared (Github Pull Request) as separate csv files containing time-resolved predictions of sample mass (file name:[Institution Name] Gasification q# Mass.csv) and back surface temperature (file name:[Institution Name] Gasification q# Temp.csv). Please follow the file formatting of template data provided on the Validation Data section of the MaCFP repository.

If sufficient agreement between model predictions and experimental measurements is not observed, pyrolysis modelers are then allowed to adjust and resubmit these parameter sets to obtain better agreement; however, they must (a) [prepare a README file identifying exactly how and why that model was changed] and (b) they CANNOT simply recalibrate their models to match this new validation data set. Adjustments to previously calibrated data sets may arrise for many reasons, including, but not limited to, changes in assumed boundary conditions of model calibration data or incorporation of new calibration data obtained in different test apparatus and/or heating conditions.

New pyrolysis models are also welcomed. Once again, we emphasize the need to use independent model calibration and model validation data sets. For this exercise, pyrolysis models should not be directly calibrated to new validation data for two reasons: (1) this would no longer make simulation results true predictions and (2) this would bias and/or eliminate the ability to objectively quantify their relative accuracy as compared to other models. Aside from this requirement, modelers are not provided limitations or suggestions regarding their pyrolysis model parameterization (i.e., calibration) approach; however, they are required to use either (a) at least one of the milligram-scale data sets (e.g., TGA or DSC) and one gram-scale experiment (e.g., cone calorimetery or controlled atmosphere gasification experiments), or (b) at least two of the gram-scale experiments available in the Calibration Data section of the MaCFP repository. Modelers can supplement MaCFP data with any literature data that they deem necessary.

Table 1 lists all pyrolysis model parameters of interest for this study. Note: degradation kinetics and thermodynamic parameters can be component- or reaction-step-specific. If your model includes multiple reaction steps and/or components, please include all relevant parameters below for each one. Participants should provide a detailed description of the method of determination of each of these parameters as well as a description (written and mathematical) of their proposed decomposition reaction mechanism.

Final submissions (Github pull request) of each new pyrolysis model (i.e., material property set) should include:

  • JSON (.json) file containing the model parameters defined in Table 1 and identifying the datasets used for calibration (please follow templates provided on the Material Properties section of the MaCFP repository);

  • Markdown (README.md) file describing model calibration approach;

  • csv file(s) containing model predictions of material thermal decomposition behavior that demonstrate proof of model calibration accuracy (i.e., if models were calibrated against mass loss rate measured in the TGA or cone calorimeter, please submit similarly formatted predictions of these datasets).

    Table 1: Pyrolysis Model Parameters

    Degradation Kinetics

    Symbol Units Name
    A s-1 Pre-exponential constant
    E J mol-1 Activation Energy
    n [-] Reaction Order
    ν [-] Stoichiometric Coefficient

    Thermodynamic Properties

    Symbol Units Name
    cp J kg-1 K-1 Heat capacity
    hr J kg-1 Heat of reaction
    ρ kg m-3 Density

    Transport Properties

    Symbol Units Name
    k W m-1 K-1 Thermal conductivity
    D m2 s-1 Mass diffusivity
    α m-1 or m2 kg-1 Absorption coefficient
    ε [-] Emissivity

Additional Guidelines for UMD-SBI

UMD-SBI (Single Burning Item) corresponds to a flame spread experiment corresponding to two (50 cm × 146 cm) PMMA plates in a comer wall configuration currently studied at the University of Maryland (see documentation here). The PMMA material is the same material studied in NIST-Gasification-Apparatus. UMD-SBI includes heat flux measurements performed both on one of the two PMMA plates and at a distance from the two plates. The PMMA material is the same material studied in MaCFP-2 and in the NIST-Gasification-Apparatus and NIST-Parallel-Panel cases. Groups interested in simulating UMD-SBI are invited to prepare simulations using two pyrolysis models (i.e., two sets of material properties): the "most average" and "most accurate" models (as defined by the condensed-phase subgroup). These property sets are available on the condensed-phase repository, matl-db; further information on how these property sets are identified is available in the Guidelines for NIST-Gasification-Apparatus section.

Suggested grid resolution: ∆x from 0.25 cm to 1 cm.

Suggested angular resolution: NΩ = 100 angles.

Plots of interest include:

  • A plot showing the time variations of the heat release rate;

  • Representative plots showing the instantaneous flame shape (e.g., identified as the 200 kW/m3 iso-contour of the volumetric heat release rate, or using some other method to be specified);

  • Plots showing the time variations of the total gauge heat flux (measured by a water cooled sensor) on the surface of one of the two PMMA plates, for different horizontal positions x, and for different elevations y;

  • Plots showing the time variations of the radiative heat flux measured at a (x,z) = (70 cm, 70 cm) horizontal distance from the corner of the two PMMA plates, in the horizontal direction facing the vertical corner, and for different elevations y;

  • Plots showing the vertical variations of the fuel mass loss rate, the surface temperature, the net surface heat flux, and the convective and radiative components of the net surface heat flux on the surface of one of the two PMMA plates, at horizontal distances (measured from the vertical corner) x = 5 and 22 cm, and at times t = 105, 145 and 185 s;

  • Plots showing the variations of the mean and rms gas temperature along the direction normal to one of the two PMMA plates, at horizontal distances (measured from the vertical corner) x = 5 and 22 cm, at vertical elevations y = 30 and 90 cm, and at times t = 105, 145 and 185 s;

  • Plots showing the variations of the mean and rms vertical flow velocity along the direction normal to one of the two PMMA plates, at horizontal distances (measured from the vertical corner) x = 5 and 22 cm, at vertical elevations y = 30 and 90 cm, and at times t = 105, 145 and 185 s.

Additional Guidelines for NIST-Parallel-Panel

NIST-Parallel-Panel corresponds to a flame spread experiment corresponding to two (61 cm × 244 cm) PMMA plates in a parallel panel configuration (featuring a separation distance of 30 cm) currently studied at NIST (see documentation here). The PMMA material is the same material studied in MaCFP-2 and in the NIST-Gasification-Apparatus and UMD-SBI cases. NIST-Parallel-Panel includes heat flux measurements performed both on one of the two PMMA plates and at a distance from the two plates.

Groups interested in simulating NIST-Parallel-Panel are invited to prepare simulations using two pyrolysis models (i.e., two sets of material properties): the "most average" and "most accurate" models (as defined by the condensed-phase subgroup). These property sets are available on the condensed-phase repository, matl-db; further information on how these property sets are identified is available in the Guidelines for NIST-Gasification-Apparatus section.

Suggested grid resolution: ∆x from 0.25 cm to 1 cm.

Suggested angular resolution: NΩ = 100 angles.

Groups are invited to first perform a simulation of the heat transfer induced by the propane burner flame without the PMMA plates. Plots of interest include:

  • A plot showing the time variations of the heat release rate;

  • Representative plots showing the instantaneous flame shape (e.g., identified as the 200 kW/m3 iso-contour of the volumetric heat release rate, or using some other method to be specified);

  • Plots showing the time variations of the total gauge heat flux (measured by a water cooled sensor) on the surface of one of the two marinite boards, at a central position (y = 0), and for different elevations z;

  • Plots showing the spatial variations of the total gauge heat flux (measured by a water cooled sensor) obtained at steady state on the surface of one of the two marinite boards, for different horizontal positions y, and for different elevations z.

Groups are invited to then perform a simulation with the PMMA plates. Plots of interest include:

  • A plot showing the time variations of the heat release rate;

  • A plot showing the time variations of the total heat flux measured as a distant location (at x = -1 m, y = -3 m, and z = 0.9 m); (note: this heat flux gauge is oriented horizontally, facing towards the gap between the two panels [i.e., pointing towards x = 0, y = -0.3 m, and z = 0.9 m])

  • Representative plots showing the instantaneous flame shape (e.g., identified as the 200 kW/m3 iso-contour of the volumetric heat release rate, or using some other method to be specified);

  • Plots showing the time variations of the total gauge heat flux (measured by a water-cooled sensor) on the surface of one of the two PMMA plates, at a central position (y = 0), and for different elevations z;

  • Similar plots showing the vertical variations of the total gauge heat flux (measured by a water-cooled sensor) on the surface of one of the two PMMA plates, at a central position (y = 0), and for different values of the heat release rate, HRR = 120, 200, 300, 400, 510, 750, 990, 1500, 1980, 2800 kW;

  • Plots showing the vertical variations of the fuel mass loss rate, the surface temperature, the net surface heat flux, and the convective and radiative components of the net surface heat flux on the surface of one of the two PMMA plates, at a central position y, and at times corresponding to different values of the heat release rate, HRR = 120, 200, 300, 400, 510, 750, 990, 1500, 1980, 2800 kW.

  • Plots showing the variations of the mean and rms gas temperature along the direction normal to one of the two PMMA plates, at a central position y, at vertical elevations z = 50, 140 and 220 cm, and at times corresponding to different values of the heat release rate, HRR = 120, 300, 750, 1500, 2800 kW.

  • Plots showing the variations of the mean and rms vertical flow velocity along the direction normal to one of the two PMMA plates, at a central position y, at vertical elevations z = 50, 140 and 220 cm, and at times corresponding to different values of the heat release rate, HRR = 120, 300, 750, 1500, 2800 kW.

Contact Information

A Google Discussion Group for the MaCFP Working Group can be accessed here. The purpose of this forum is to help develop the network between fire researchers scientists, provide a community-wide forum for discussion and exchange of information, and to facilitate data sharing and model development to improve computational predictions of fire behavior.

For questions, please use the following points of contact:

MaCFP-3 Session/Case Session Co-Chairs Technical Point of Contact
NIST-Waterloo-Pool-Fires Tarek Beji, Ryan Falkenstein-Smith Anthony Hamins, Beth Weckman
FM-Burner Ning Ren, Gang Xiong Yi Wang
NIST-Gasification-Apparatus Jason Floyd, Bjarne Husted Isaac Leventon
UMD-SBI Dushyant Chaudari, Alexander Snegirev Stanislav Stoliarov
NIST-Parallel-Panel Lukas Arnold, Kevin McGrattan Isaac Leventon

Please cc Arnaud Trouvé, Bart Merci, Isaac Leventon and Randy McDermott with questions to help ensure prompt responses. Thanks!