New Performance-Based Method for External Fire Spread Assessments and Other New Software for Fire Protection Engineers
By Piotr Tofiło and Wojciech Węgrzynski
The magnitude of radiative heat flux is the key factor in assessing the likelihood of external fire spread between buildings or in different parts of the same building, as the separation distances required by regulations depend on the extent of critical heat flux in the space around neighboring buildings.
Fire regulations and standards [1–3] typically address only simple geometries based on a radiating rectangle model, which can be calculated relatively simply by hand. This is, however, not always optimal for real projects featuring more complex architectural configurations, which may include irregularly spaced and oriented radiators, angular façade geometries or intervening objects that act as obstruction for heat radiation. Some standards such as Eurocode  go further and describe the geometry of external flames, which should be included to make the assessment more accurate and conservative.
All this may lead to heat flux distribution that is difficult to calculate, and the tools to do this are not widely available. Of course some people may resort to computational fluid dynamics (CFD), but this approach has some difficulties, too. In CFD the engineer must adequately model a fully developed fire, which is a challenge in itself due to many modeling uncertainties like fuel and material properties, more complex chemistry, underventilation and so forth. These conditions are still considered extreme and challenging for CFD packages. Of course, CFD will output some figures and visualizations, but a critical engineer must understand the scientific quality of this output. There is significant concern that unexperienced CFD users may produce unrealistic fires, and therefore, it is safer for them to stick to temperatures (radiator/flame) advised by regulations and standards.
Another problem lays in radiation models implemented in CFD codes, which often involve intensive calculations. Unfortunately, simple tests reveal numerical weaknesses leading to unrealistic heat flux distributions for insufficient spatial and angular resolutions in popular CFD packages (see Figure 1). At best, the engineer must accept the fact that precise radiation calculations increase computation time significantly, not to mention a long duration of a fully developed fire. Another aspect that may be an obstacle is rectangular meshing, which may not resolve properly parts of façades that are not perpendicular in shape. These drawbacks make CFD currently rather unpractical for a quick heat flux assessment involving several geometrical configurations.
Some other aspects to consider during heat flux assessments is the local orientation of the exposed surface when dealing with flammable building elements attached to the wall surface (e.g., cables, plastic pipes, building interiors close to windows). Many engineers assume that the local orientation vector is always normal to the surface, which is not always the case. In some applications a higher level of conservatism may be required. It is worth noting that CFD can output the heat flux only for a specified orientation, typically the orientation of a solid wall.
As a solution to the calculation difficulties discussed above, the FireRad software program was developed. The main ideas behind FireRad is that it should be accurate and very quick to use. The user should be able to prepare the building model by importing an existing 3D model or interactively from existing elements, which takes very little time. Calculations in the simplest resolution should not take longer than few seconds. The theoretical algorithm consists of view factor computations based on triangles, as they are the most basic shape in geometry, and every complex shape in 3D can be tessellated into triangles.
The basic shapes available in the program include several flat surface types (radiating, receiving, obstructing, obstructing/receiving), several 3D objects (external radiating flame with thermal gradient, radiating cone, cylinder) and a point source as the simplest form of a radiating object (Figure 2). The user can specify the properties of each object interactively (e.g., by adjusting dimensions, location, temperature, heat flux, emissivity). The external flame with thermal gradient is based on Eurocode procedure, so to use such an object, the user must define the fire compartment (size, openings, fire load density) to establish flame size, compartment temperature and flame temperature. The more complex objects have several additional parameters such as deformation or number of sectional divisions. The assisting shape (Box 3D) can be used to quickly build the basic geometry of the building from several boxes of different shapes, sizes and rotations. Holes and recesses can be quickly created using Boolean algebra. Every object can be positioned, moved, rotated and copied into a matrix.
The user can see results in several ways. This can be 3D, 2D and 1D distributions (Figures 3 and 4). The user decides the calculation grid size (10, 20, 50 cm) and the orientation of measurement points. Calculation grid size, unlike in CFD, does not affect accuracy at each measurement point, but it uses a denser matrix of points to improve the visual quality of 2D or 3D output. This means that the initial calculation can be done quickly on the coarsest grid. All the results can be saved as an Excel spreadsheet for further analysis or processing.
By default, the orientation of all measurement points is normal to the surface, but the user can choose other vectors or an adaptive orientation. Adaptive orientation means that the orientation vector of the measurement point is automatically adjusted to achieve the highest heat flux from all radiating surfaces seen from that point. This mode is created for a higher level of conservatism to address the uncertainty related to the local orientation of the receiving surface. This treatment affects areas that are at bigger angles to radiating surfaces. There is no change in heat flux in areas where the vector facing the center of the radiator is at a right angle. The difference between the heat flux distribution for fixed vectors and adaptive orientation can be seen on Figure 4. The extent of critical heat flux is noticeably wider for adaptive orientation, because in these areas heat flux is much lower using vectors perpendicular to radiator rather than vectors targeting its center point.
Ability to use obstructions is an important feature of the FireRad. Every polygon defined as an obstruction is obstructing the passage of radiation to a carefully calculated degree. It is implemented by projecting each obstructing polygon on all radiators from all measurement points. If an obstructing radiator creates a shading, then the radiating polygon has to be modified by removing the shaded area. The remaining part of the radiator is tessellated into triangles. This concept is presented in Figure 5.
As was already mentioned, the important feature of the FireRad is the importing of existing models from CAD or BIM software (e.g., Revit) using a universal STL format. First, one or both buildings are imported and positioned in a desired way, and in the second step the user points and defines the areas of interest, which are important in calculations. The whole process is very quick, and the results of calculations can be seen within seconds, as shown in Figure 6.
FireRad can be also used in ways that are less established in regulations such as radiation from windows and unprotected areas, open roof fires or pool fires. An example of two neighboring buildings is presented in Figures 7 and 8. The small one-story building has an unprotected flat roof. To analyze the need for separation of two buildings, a prismatic flame zone was modelled over a flat roof to the height of 1.4 stories as recommended by NFPA 80A . The surface of the flame zone is assumed conservatively to be 900 °C, and emissivity is 1 based on Babrauskas .
Three wind-deformed shapes of flame zone were analyzed to study the impact on the neighboring building. Critical heat flux on the façade was exceeded only in the most unfavorable wind case. As a result, these areas need to be of adequate fire resistance. An additional more detailed study was performed to establish whether the walls of the recessed balcony need protection. The heat flux distribution in Figure 9 shows that the critical heat flux is not exceeded in these areas. Note that for this additional measurement the column and both side walls were defined as both receiving and obstructing.
FireRad has been thoroughly verified and validated. This gives a good computational framework, but some assumptions about temperatures and emissivities are necessary to use it with safe conservatism. FireRad results were also compared with recent comprehensive experimental studies performed in Canada. The comparison showed that the use of external flames was necessary to obtain good agreement. The paper discussing the methodology of calculations in greater detail has been recently submitted to Fire Technology .
Members of the Polish Chapter of SFPE were also involved in developing other fire-engineering software. One is QuickZone, which is essentially an interactive zone model using the Consolidated Model of Smoke and Fire Transport (CFAST) that allows quick model building and immediate monitoring of results. The idea is to enable the engineer to quickly create the model and see all results in the same screen without the use of spreadsheets. Such quick investigation is sometimes desired in the fire engineer’s workflow and also brings the opportunity for more convenient parameter testing or sensitivity study. It also facilitates general understanding of compartment fire dynamics, which is especially valuable for less-experienced engineers, who can quickly see how different parameters affect fire dynamics and fire environment. Further works are currently under way to include various aspects of probabilistic analysis in QuickZone.
Another software developed recently is FDS Designer, which is a graphical interface for FDS. The concept of such software is not new, but it has some new interesting features. The most important one is seamless integration with cloud-based computational resources, FDS Cloud. The user can conveniently run simulations without the need for clusters or work stations. The user selects the size of the virtual machine (number of cores and amount of memory).
Piotr Tofiło and Wojciech Węgrzynski are with SFPE Poland
FireRad, QuickZone and FDS Designer with FDS Cloud are modules of a single software, Fire Engineering Platform, which is available at www.fireplatform.eu. This site also gives access to FDS Cloud resources (also accessible from www.fdscloud.eu), which can be used from the website without any software. This is a convenient, quick way to upload an FDS file and run simulations from any device on the web.
Note that all programs are at an early beta-testing stage. Minor bugs may be encountered. No charge is necessary, but the user must obtain a license key to use the package. Any changes to this policy will depend on the size of the user community and needs for maintenance and further development, as several new ideas are being considered for new modules to enhance the platform and create a versatile tool for fire engineers.
Approved Document B (Fire Safety), Vol. 2, Buildings Other Than Dwelling Houses, 2013.
- External Fire Spread: Building Separation and Boundary Distances (BR 187, 2nd ed.), 2014.
- NFPA 80A: Recommended Practice for Protection of Buildings From Exterior Fire Exposures, 2012.
- EN 1991-1-2: Eurocode 1: Actions on Structures, Parts 1–2: General Actions – Actions on Structures Exposed to Fire, 2002.
- V. Babrauskas, Temperatures in Flames and Fires (San Diego, CA: Fire Science and Technology, 2006). http://www.doctorfire.com/flametmp.html.
- P. Tofilo, and V. Mozer, Radiative Heat Flux Calculations for Complex Geometries, submitted to Fire Technology.