Fire Safety Engineering: Applications for Water-Based Fire Control and Suppression Modelling
By Grégoire Pianet, Alexandre Jenft and Armelle Muller
Presently, many applications of computational science to fire safety engineering (FSE) are directed at heat and smoke propagation in buildings and thermal stress on structures, as well as the control of each by means of specific fire protection systems (smoke control systems, fire walls, smoke curtains, etc.). There are two reasons for this: on one hand, both hot gas propagation models and solid structure stability models are reliable and can be solved with enough accuracy for engineering applications; on the other hand, the calculation power of both scientific and personal computers has dramatically increased over the last 20 years. As such, ordinary computers can handle calculations of aeraulic phenomena or structure stability for standard sized buildings.
Furthermore, fire protection in buildings also includes active systems like sprinklers or water-mist which are designed to suppress fire, control fire propagation, and cool hot gases and smoke. As water-based systems are known to strongly affect smoke propagation and thermal stresses on structures, both contracting and civil authorities are justified in demanding evaluations of water-based protection systems efficiency when considering an FSE project.
Fire control and suppression phenomena
This is a far more challenging field since modelling fire and water interaction involves a multiphase medium compound of hot gases, soot, water droplets and water vapor, interacting with either burning, combustible or inert solid surfaces. One of the most important issues involves the modelling of fire behavior under aspersion which is driven by several phenomena:
- Heat absorption by vaporization (including hot gases cooling, solid surface cooling and pre-wetting);
- Inerting by oxygen dilution;
- Radiative heat flux attenuation.
Considered independently, these physical phenomena have already been modelled with success. But real-size fires with water-based protection involve a strong coupling of those phenomena where modelling still needs to be addressed by research efforts.
Fire control and suppression models
One can distinguish three families of models which have already been used to evaluate fire and water spray interactions:
- Commercial multiphysics codes, provided with aspersion and cooling models, not specifically developed for FSE applications but used as such;
- High-end computing codes, with specific fire and combustion toolboxes, but whose computational cost in size and time makes it almost unusable for real-life engineering applications; and
- Very fast codes specifically developed for FSE, but with strong hypothesis and numerical parameters settings which cannot be defined in a predictive way.
The first family could eventually be suited for applications with little or no coupling between water-spray and fire, as it is not designed for FSE, it generally suffers from a lack of verification and validation. Codes from the second family are not suited to FSE studies because of very demanding computational resources in terms of both size and time, and, as such, are better considered as powerful research tools for building intermediate models.
Amongst codes from the third family, the Fire Dynamic Simulator (FDS) from the National Institute of Standards and Technology is used worldwide by the FSE community. Models of water aspersion were included several years ago, but only those involving empirical parameters, which cannot be predicted, which makes the model almost applicable for engineering applications. For inexperienced operators, using default values for these parameters may produce a convincing result, but it would be subject to strong error levels.
The authors believe that technical centers specializing in fire suppression should play a role in, through their "duty to advise" in communicating the limitations of engineering computing codes and in helping researchers develop new engineering models based on a large experimental knowledge. This is the point of research activities (see [1, 2]) funded by CNPP Group with LEMTA Laboratory (UMR 7563).
Experimental and numerical modelling project
An exhaustive experimental campaign has been realized with a goal of better understanding what mechanisms drive interactions between fire and water at a laboratory scale, but still using conventional aspersion systems. In addition to being videotaped, instrumentation monitored gas temperature, combustible surface temperature, oxygen consumption, pyrolysis rate and thermal flux. Over 80 fire tests based on liquid pool fires were conducted with different input parameters such as combustible type, heat release rate, number of active nozzles and time from fire start to aspersion.
The objective was to identify suppression mechanisms and to discriminate physical parameters to be used in a predictive model. We identified two mechanisms: suppression by fuel cooling (as fuel temperature is lowered below its ignition point) and suppression by inerting effects (as water vaporization causes oxygen dilution yielding suppression by lack of combustion air).
In this project, research efforts have been focused on the relation between heat release rate and fuel surface temperature after water aspersion. The Arrhenius law, usually considered to model rate of pyrolysis before water application, was modified providing a new model based on fuel temperature, fuel ignition temperature, and two numerical parameters. Unlike the model currently implemented in FDS code, it was found that the two numerical parameters can be predicted if fuel behavior before water application is well known. As such, parameters can be determined using simulations of the free-burning phase.
Promising results were achieved, as the new model predicted all suppression cases by fuel cooling. In one case, the fire was not suppressed which was confirmed by simulation. Suppression time was estimated within an acceptable order of magnitude considering FSE applications generally underestimate suppression time due to overestimation of heat exchange between water drops and solid surfaces. The current objective is to improve suppression time prediction, and apply the model at different scales and fuel configurations.
Remaining issues that still require R&D
The research project helped identify issues in the model that are currently implemented in FDS and other codes. The research project has focused on fuel cooling but it is common knowledge that local suppression by inerting and generally by simulation of under-ventilated conditions needs significant improvement (see mixture fraction model in FDS). It is also admitted that the evaporation model is quite efficient for smoke cooling but heat absorption is still over-predicted, with error levels that tend to increase as heat absorption occurs in flaming zones.
Large scale modelling attempts
With a goal of testing the maturity of current water-based suppression models for FSE applications, simulations of past large-scale fire tests were performed.
1.7 MW pool fire in engine room with water-mist application. Suppression in tests and simulations were observed. As previously stated, there is a significant trend to underestimate suppression time due to overestimated exchanges between water drop and solid surfaces.
0.4 MW then 5 MW pool fire in aircraft hangar with water-mist application. There was no suppression observed in tests where a very slow suppression was observed in simulation. Once again, this was ascribed to overestimated surface/drops heat exchanges. However, a large uncertainty level due to nozzle type and spray PSD have been shown to be parameters which have significant impact on simulation sensitivity.
0.8 MW fire in hotel room with water-mist application. Suppression in fire tests and simulations were observed within comparable delays. It was found that combustion zones persist for screened surfaces (under bed) in simulations.
One can see that results on large-scale configurations are qualitatively acceptable when used and interpreted together with fire test data. It was not possible to obtain a quantitative agreement for several reasons:
- Models of gas and solid surface cooling still need improvements;
- Metrology of fire tests for large-scale configurations was not designed for simulation needs (fuel surface temperature);
- Some data in these experiments were only partial (particle size, distribution of nozzles, pressure in pipe-networks, etc.) which had a significant impact on simulations.
Application perspectives to FSE
This section attempts to outline what FSE applications of water aspersion modelling could be expected in the near future. Our research revealed that crude modeling of a water aspersion system with little or no specific experimental background yields large error levels.
Nevertheless, FSE applications can support a certain error level depending on the provided methodology. The dimensioning approach, for example, predicts suppression time or no suppression at a given accuracy. A security-oriented approach could consist of evaluating whether a water-based protection system is able to control a developing fire. Irrespective of the time it takes to achieve the goal, an answer to the latter question is potentially conclusive from an engineering perspective. A relative or comparative approach could consist of stating which water-based protection system is best suited to a particular application. If all relevant physical phenomena are taken into account, the relative approach is known to be reliable. Therefore, application perspectives depend on one of three conditions:
- The physical setup on which the code is applied has no or little use of the weakest models;
- The weakest models are improved or replaced by more efficient ones;
- Numerical model is systematically validated by an experimental setup with a proper metrology. The following table summarizes what application and methodology could be used together considering current water-based control and extinction models. For some combinations, specific care is required which means numerical modelling should include specific verification, validation and sensitivity analysis, together with a significant background in suppression tests.
A new predictive model of surface cooling developed for FDS has proven effective with medium scale tests. Model parameters for a well defined material can be determined using simulations of the free-burning phase, or using results from specific fire tests with adapted metrology. Besides this significant progress should now focus on improving drop/wall heat exchanges models, inerting models and evaporation models.
Using engineering-oriented computing codes for water protection design is currently possible in very few applications. Extending the application range in the near future is possible but still subject to research and development. Further collaboration is needed between academic laboratories and technical centers that have real-life experience with specialized fire extinction design that are able to provide relevant facilities to ensure assessment of numerical models from medium to real-size scales.
Grégoire Pianet, Alexandre Jenft and Armelle Muller are with Groupe CNPP - Fire & Environment Department
- A. Jenft, P. Boulet, A. Collin, G. Pianet, A. Breton, A. Muller. Can we predict fire extinction by water mist with FDS? Mechanics & Industry; 14(5):389-393, 12/2013. DOI: 10.1051/meca/2013079
- A. Jenft, A. Collin, P. Boulet, G. Pianet, A. Breton, A. Muller. Experimental and numerical study of pool fire suppression using water mist. Fire Safety Journal; 07/2014; 67:1–12. DOI: 10.1016/j.firesaf.2014.05.003
A. Jenft. Study of the interactions between fire and water mist systems. Development of a suppression model for FDS software, PhD Thesis Université de Lorraine, 12/2013.