RICERCA ITALIANA     

Prediction of thermo-fluid-dynamic and structural effects of tunnel fires, for risk analysis and emergency management

Università degli Studi di Padova

Abstract

Since a long time tunnels are very attractive for easier communications (by road or rail) in presence of natural obstacles (such as mountains and rivers) or even existing urban areas. Not surprisingly, Italy with its peculiar orographic conformation and geographic positioning, has the highest European tunnel lenght (more than 1900 km, not including city metro tunnels), equal to 27% of the European total. However, in parallel with an always increasing interest for underground transportation structures, there is also an increasing concern for possible fire emergencies occurring inside tunnels, taking into account the peculiar geometry of such infrastructures and the intrinsic risk of many freights.
Therefore, continuing a research activity started in 2004 within a PRIN Project, this new research Project aimed to maintain the effective links established among several Italian research groups which are very active in different fields related to tunnel fires, ranging from combustion to fluid-dynamics, from thermo-structural problems to construction technologies and ventilation systems.
The strict cooperation among these groups, each bringing its know-how, will allow not only for studying all the different aspects the fire scenario, strictly interrelated to each other, but especially for developing an unitary approach for risk assessment and safety planning, as requested by designers, traffic managers and local authorities. <<<

Principal Investigator

Pierfrancesco Brunello Università degli Studi di PADOVA

Research Objectives

As already mentioned, this Research Project will be addressed to proceed in the research field already dealt with during previous activities financed by MIUR for the years 2004-2006, with the scientific coordination of prof. P. Brunello.
In this regard, several changes have been also introduced, both in the Units participating to the Project, and in the researchers involved; thus, specific competences have been added, especially in the fields of electrical and ventilation systems and in the area of risk assessment and emergency management.
The main activities will be briefly described hereafter, but the reader will certainly find more information in the B Models of the various Research Units.
From the very beginning of the biennal activity, the various numerical models will be upgraded and accompanied by suitable experimental measurements, in order to provide the necessary input data and also to validate the numerical results against experimental data.
Starting from combustion phenomena, the Unit of TRENTO (prof. Baggio) is planning to gain a better knowledge of the behavior of materials during a fire both trough experimental investigations and by extending the capabilities of the numerical model previously developed. Such aim will be pursued by experimental analyses of pyrolysis taking place during the heating phase of materials affected by open fire: this activity will be carried out by heating the materials according to a preprogrammed temperature ramp inside a custom built apparatus. Then, pyrolysis will be described by means of a thermodynamic equilibrium model of the chemical reactions, to be validated by comparison with experimental results; this way, also the development of a simplified combustion model (solid phase -gas phase) will be possible. Finally, the aforementioned model will be integrated into an existing heat and mass transfer model for the simulation of the combustion chemical reaction (able to estimate the local release of thermal energy due to reaction and the local rate of reactant conversion into products. Also the reliability of the complete combustion model will be assessed by means of experimental data.
As far as fluid-dynamics is concerned, this field will be mainly covered by the Unit of TORINO (prof. Borchiellini) with the main purpose op upgradin the existing “zone model” developed during PRIN 2004. This model is certainly most suitable for analising a large number of fire scenarios; therefore, he objective of this phase will consist in evaluating the possible advantages in terms of accuracy in the results that can be obtained by introducing a third zone representing air and smoke mixing. In this activity a more detailed model will be assumed as a reference, with a CFD approach close to the fire and a one-dimensional model to describe the rest of the tunnel. It is important to remark that simulations will be carried out in transient conditions, defining the air pressures at the tunnel portals, the location of fire and the power profile.
To fluid-dynamics an important contribution will be provided also by another Unit of PADOVA (prof. Brunello). Taking advantage from previous experiences of some researchers belonging to the Unit in the fields of mechanical ventilation of road tunnels, the numerical modeling of fans will be improved.
First, an experimental campaign will be carried out on a scale model road tunnel, already available at the Department of "Mechanical Engineering" of Padova University. The experimental apparatus will based on grid of pressure sensors, installed at regular intervals on the surfaces of the model tunnel, in order to analyse the pressure fields upstream and downstream the ventilation fans. These pressure fields will be used to determine the optimal distance between fan installations as a function of the average operative condition in the tunnel. The installation efficiency will be also investigated by Pitot aerodynamic probes to measure the velocity profile downstream the ventilation fans.
After the experimental campaign, a series of three-dimensional CFD simulations will be performed using a commercial code. With these simulations the entire flow field, the transport of gases (e.g. pollutant, smoke), and the heat transfer inside the tunnel will be analysed. The numerical domain will enclose a portion of the road tunnel around a single fan or a side-by-side pair of fans. The experimental profile of velocity and pressure will be used to impose the boundary condition of the computational domain.
Using the experimental data, a critical evaluation of the performance of various turbulence models will be performed in order to assess which of them provides the best prediction of the complexities arising from the stagnation regions, diffusion, and other three-dimensional effects. <<<

Timescale

24 months

National and international background

It is well known that for obtaining reliable analyses of tunnel fires, various phenomena related to different research fields have to be considered at the same time.
Traditionally, fluid-dynamics is certainly dominant when tunnel fires are analysed, because of the strong effects of fire on temperature and nature of the fluids involved (fresh air and combustion products). This situation is also due to the fact that fluid-dynamics is important also for the simulation of tunnels under normal operative conditions, since the evaluation of the air quality is often required.
For this purpose classical semi-empirical relationships, based on the energy and mass balance of ducts, have been used for a long time. Some applications of this kind have been proposed also by international institutions, for instance the “Centre d’Études des Tunnels (CETU)” in France.
Recently, since computational resources drastically improved, the possibilities offered by the so-called CFD (Computational Fluid Dynamics) have been often emphasized. In fact, nowadays several CFD codes are available, both for general purposes and specifically for fluid-dynamic analyses of tunnels (also during fire). One of the most important dedicated software is SOLVENT, developed by ASHRAE in the framework of the “Memorial Tunnel Fire Ventilation Test Program” [1].
However, in spite of the above mentioned improvements of computational capabilities, CFD analyses still require considerable resources, especially if large space domains and/or long time spans must be considered. Unfortunately this is the case of tunnel fires, when a large number of numerical simulations are ususally necessary to analyse several different scenarios and design solutions.
Therefore, the so-called “zone models” [2], initially developed for fires in buildings [3] and characterized by an intermediate complexity level, have been proposed also for tunnel fires.
The “zone modeling” approach has been widely applied also by one of the Units involved in this Research Project (prof. Borchiellini): starting from an existing one-dimensional model [4], [5], [6], their “zone model” is based on a hybrid approach: two zones are used in proximity of fire (where smoke and fresh air are stratified), while a one-dimensional approach is applied for the rest of the tunnel and for the ventilation ducts. Therefore, this model can be used for extensive simulations on actual tunnels (with a complex geometry and/or particularly long) and has been succesfully applied also for analysing some aspects of the Mont Blanc incident [7]. The same concept can be applied also to obtain more sophisticated hybrid models, by using detailed CFD modelling for the zone close to the fire, whereas a one-dimensional analysis of the rest of the tunnel provides the boundary conditions to the CFD code [8].
In this Research Project the activities related to fluid-dynamic simulations will be accompanied also by experimental tests on a scale model of a tunnel. This model has been set up and instrumented by researchers belonging to one of the Units of Padova University (Prof. Brunello), which have a broad theoretical and experimental background in the field of motorway tunnel ventilation [9], [10].
However, while fluid-dynamic aspects of tunnel fires are often treated with a satisfactory level of accuracy, some other aspects of the global phenomenon are drastically oversimplified or even neglected.
For instance, this is the case of combustion phenomena, which is the very basis of the fire. This aspect is usually taken into account simply by means a suitable heat generation inside the fluid. No doubts that the dynamics of combustion in tunnel fires is much more complex than in controlled laboratory conditions: for these situations reliable numerical models are available since many years. Instead, during a real fire in tunnel the quantity and quality of flammable materials can widely change, while the evolution of the fire and the type of the combustion products are the result of the interaction with mass flow rates and heat transport phenomena between the reaction zone and the surroundings.
In the framework of the previous PRIN 2004, two Units (one of them, lead by prof. Baggio) were focused on theses topics and the problem has been analysed from a theoretical and experimental point of view. Since reactions take place in a heterogeneous phase (i.e. liquid-gas and solid-gas), the modelling of combustion requires the modelling of two phenomena which are strictly related but substantially different: the formation of volatile products (also by pyrolysis) and the subsequent combustion in gaseous phase.
Combustion of volatile products combustion can be modeled independently on the substance which released the inflammable gases, while pyrolysis is strongly dependent on the type of the original material, in term of both evolution and final products.
Therefore, taking into account some recent results obtained in this field [11], [12], the above mentioned two Units focused their research activity to the experimental analysis of the pyrolysis phenomenon for the combustible substances present in tunnels (due to vehicles and to various tunnel elements) and to the development of a theoretical model based on a simplified version of the equations necessary for describing the pyrolysis and gaseous processes which take place inside chemically reacting materials.
For the combustion of pyrolysis gases it possible to use a simplified description of the phenomena by considering the amount of oxygen inside the tunnel and the bulk temperature of the gaseous phase. In such a way the average temperature of the flame, its extension and the quantities of the main gaseous components originated from the reactions, can be easily obtained.
Finally, the direct combustion of the solid carbon residuum can be treated in a simplified manner as a function of the oxygen quantity reaching its surface.
Therefore, the analysis of the combustion and of the fluid-dynamics inside the tunnel are strictly connected and they are essential parts of a complete model for fire simulations.
Another aspect which is very seldom considered in most models currently applied for fire simulations, is the radiative heat transfer which takes place in the tunnel during an accident. Usually, only the convective aspects are considered, by assuming that the energy released by the chemical reactions is entirely addressed to the “plume” of smoke. On the opposit, it is an experimental evidence that also the radiative heat exchanges are of great importance, since they can significantly affect the thermal field in the tunnel: in fact, during a fire many surfaces have high temperatures and emissivities and additionally several strongly participating media (the products of combustion) are involved in the radiative processes.
In the framework of the previous PRIN 2004, part of the research activity has been addressed to these aspects. An existing numerical model, based on a stocastic Monte-Carlo algorithm and already successfully applied in different fields (from high temperature heating panels [13] to baffling systems of space vehicles [14], from solar pressure evaluation on satellite antennas [15] to daylighting in buildings [16]) has been improved and linked to a CFD model to simulate the radiative behaviour of a tunnel fire.
Even less frequent in current fire models is the analysis of the actual behaviour of the structural elements inside the tunnel, in spite of their importance in terms of damages and human losses. In fact, these elements can be made of combustible material (e.g. asphalt contained in the road pavement) or they can collapse, with strong consequences for the catastrophic effects of the fire event. In this regard, the seawater infiltration observed after the fire in Channel Tunnel or the spalling of vault portions in Tauern Tunnel (wide enoug to prevent the access of rescuers), can be mentioned.
Researchers belonging to two Units of Padova (proff. Brunello and Pesavento) and to the Unit of Trento (prof. Baggio), have gained a good background in the field of heat and mass transfer in porous media, considering the thermo-mechanical aspects of the problems. A new model for the analysis of heat and mass transport phenomena in porous media has been developed [18], [19]; lately, it has been extended in order to consider stress-strain state takin place in the materials, first at ambient temperature [20], [21] and then at high temperature [22], [23]. Recently, the model has been further improved for taking into account the multiphase porous nature of concrete [24] and the behaviour of water above its critical point [25], as well the so-called “Load Induced Thermal Strain (LITS)”, the micro-cracking and the thermo-chemical deterioration of the material [26]. The model has been successfully validated against published results of several experimental available in the literature [27], [28].
The various versions of the abovementioned model require a deep knowledge of some thermophysical properties of the materials involved: in this regard, some researchers of the Unit of Padova University (prof. Brunello) have a wide experience at international level in the field of experimental measurements of thermophysical properties of porous materials, both in wet and dry conditions. In this field not only the measurement techniques of thermal conductivity of dry porous media have been improved [29], but also a crucial contribution in the definition of the main physical quantity for coupled heat and mass transfer (the “hygro-thermal trasmissivity”) has been given: now this quantity is defined also at international standard level by ISO 10051 "Thermal insulation - Moisture effect on heat transfer - Determination of hygro-thermal transmissivity". During the last years research efforts have been addressed on this property, also in the framework of two other PRIN Projects (1999 and 2001) [30]. Afterwards, during PRIN 2004, the researchers focused their attention to the measurements of thermal conductivity at high temperature. For this purpose, the existing apparatus available in the laboratory of the Department (based on the hot plate method and on the use of thermal flow-meters) have been modified and a suitable new one, based on the hot wire method, has been identified. The new apparatus can be used in transient conditions and at high temperature for structural materials with medium to high conductivity (not compatible with existing devices), which are typical for tunnels.
Finally, it is worth to mention that in the previous PRIN 2004 many efforts were devoted to a suitable linkage among the numerical models proposed by the various Units: this research activity started some years ago within a cooperation between two Units of Padova University (prof. Brunello e prof. Pesavento) [31].
As already mentioned, this new Project will be addressed also to new topics, strictly related to the previous ones, and suitable numerical models for the simulation of the fire will be applied for risk assessment and for proper emergency management. In this field some important activities have been already carried out both in Italy and abroad: for instance, recently in the United States some guidelines for the analysis of the risk in the transportation of chemical substances have been proposed [32]. Similar initiatives have been proposed also in the United Kingdom [33] and in the Netherlands [34].
In particular, as far as the safety in tunnels is concerned, the decisions are directly connected with the evaluation of the costs of safety measures vs. the benefits deriving from a reduction of the hazards.
Therefore, the decision-making process requires the knowledge of the potential consequences of an accident in tunnel in terms of human losses, structural damages and economic costs. Many researches have been developed in view of the definition of a quantitative methodology for the assessment of the consequences of the incidents in tunnel [35], [36]. In recent times, a dynamic model taking into account both the physical effects of the fire and the rescue possibilities of the people involved, has been developed [37]. This model, based on the analysis of the accidents occurred in the past and on criteria available in the literature, allowed for a detailed analysis of various elements and to find out some interesting correlations.
As far as the Italian situation is concerned, an interesting research in the field of hazardous material transport has been developed in the framework of the research Project promoted by the “National Group for Defense against Chemical, Industrial and Environmental Hazards”. In this area some codes for the assessment of the risk in the transport system have been already developed and in this field researchers of the Unit of Padova University (prof. Maschio) gave an important contribution, as far as the definition of innovative procedures for the evaluation of the risk level is concerned, as a function of the transport media [38],[39]. <<<