Available online at ScienceDirect. Procedia Engineering 192 (2017 )

Transcription

1 Available online at ScienceDirect Procedia Engineering 192 (2017 ) TRANSCOM 2017: International scientific conference on sustainable, modern and safe transport System for deterministic risk assessment in road tunnels Ondrej Pribyl a, Pavel Pribyl a, Tomas Horak a * a Czech Technical University in Prague, Faculty of Transportation Sciences, Konviktska 20, Praha 1, , Czech Republic Abstract In this paper, a pragmatic and goal oriented system for risk analysis in road tunnels is described. It is particularly focusing on mortality risks in case of a road tunnel accident. It is a deterministic approach combining three major components: a) vehicle distribution in a tunnel; b) smoke propagation in case of a fire; and c) people evacuation (escape) component. The major improvement of this approach is in capturing the knowledge often provided only by experts into a robust and pragmatic system available to all decision makers. This is achieved through a large number of scenarios combining different configurations of road tunnels (e.g. different number of lanes, different speed limits) and the travel demand (e.g. different structure of the flow, different volumes of traffic) which were prepared and evaluated through microscopic traffic simulation. The resulting scenarios with the information about the number of vehicles in different tunnel sections were obtained. The results describe most of the existing tunnels and situations and can be used universally. Similarly, the people evacuation component can be evaluated in a general form. The results can be manually updated to suit any particular road tunnel which can differ for example by the availability or quality of warning and information systems. The remaining task is to create a physical model of the real tunnel and to model the smoke and fire propagation. All these components were combined into the CAPITA software that was developed as a part of the research project HADES (supported by the Technology Agency of the Czech Republic). The CAPITA software is presented in the last chapter of this paper. The scenarios available off-line are in fact a knowledge base available to experts as well as decision makers and lead to a higher level of comprehension of the developments in case of fire and significantly speed up preparation of a risk analysis Published The Authors. by Elsevier Published Ltd. by This Elsevier is an open Ltd. access article under the CC BY-NC-ND license Peer-review ( under responsibility of the scientific committee of TRANSCOM 2017: International scientific conference on sustainable, Peer-review under modern responsibility and safe transport. of the scientific committee of TRANSCOM 2017: International scientific conference on sustainable, modern and safe transport Keywords: road tunnel; risk analysis; deterministic model; CAPITA * Corresponding author. Tel.: ; fax: address: horaktom@fd.cvut.cz Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license ( Peer-review under responsibility of the scientific committee of TRANSCOM 2017: International scientific conference on sustainable, modern and safe transport doi: /j.proeng

2 Ondrej Pribyl et al. / Procedia Engineering 192 ( 2017 ) Introduction of providing adequate level of safety for the end user, in this case passing vehicles. There is always a moral dilemma on the side of the engineers and decision makers in general to find a compromise between the installed safety equipment and a safety level of a tunnel [1]. It is not possible to equip a tunnel with a safety equipment or a system to make it absolutely safe. On the other hand, installation of each additional safety equipment or system that is not needed to maintain a specific safety level, brings higher investment costs as well as (often disregarded) operational costs. The design of the technological safety system is based on technical standards. There is a perception that a tunnel is safe when it is designed according to these technical standards. Such approach has certain limitations, since the tunnels are seldom identical, each tunnel has some unique features [2]. Even though the tunnel fulfils all requirements set in the technical standards, there is a residual risk that can be never specifically addressed. In many cases the real situation is different from the anticipations of the authors of the technical standards, which are often prepared over a lengthy period measured in years. This means that even when the technical standards are thoroughly applied, there is always a residual risk that cannot often be named or even known. formalistic approach may easily lead to increased construction costs. The general discussion on how to properly equip organized by the World Road Association PIARC it is possible to say that there is an increasing agreement on the need to optimize and lower costs of the equipment installed in tunnels, which leads to the approach combining the application of technical standards with risk analysis. The following paragraphs briefly introduce the risk analysis model, which is available to the larger professional audience and that can be used along with the current prescriptive approach to tunnel design. 2. System approach to risk analysis The Ministry of Transport of the Czech Republic and the Road and the Motorway Directorate of the Czech Republic paid much attention to research aimed at optimization of the safety equipment in road tunnels, which lead to a more holistic view of the safety standards covering not only tunnel design, but also tunnel operation. Risk assessment differs throughout the documentation preparation process depending on the stage that is currently elaborated [3] Qualitative methods of risk assessment Qualitative methods of risk assessment are methods that directly express the level of risk in the form of a numerical value indicating the number of affected people, economical losses, etc. Generally speaking, these methods are the most objective ones, since they are relatively well manageable. They are divided into methods working with statistics, which give us as an output in the form of a probability of a certain risk (for example death of people), and into the methods using transport simulation and physical m number of killed people. The key factor in a modern risk assessment is understanding that undesirable occurrences do not happen by themselves, but that they start with (often negligible) disruption of the normal operation that can unfold into several directions based on the reaction to the initial occurrence. This concept is known as the Bowtie Model and is shown in Fig. 1. The upper part of the figure shows the principle of the probabilistic method and the lower part of the figure shows the principle of the deterministic method called the Scenario Analysis Method, which is described later on as the software CAPITA. The left side of the model is dealing with the cause of the undesirable occurrence such as fire, accident, traffic congestion or loss of freight. The occurrence, resp. estimation of its probability or frequency, is usually done using the Fault Tree Analysis (FTA). The FTA uses inputs such as traffic intensity, vehicle fleet composition, tunnel construction parameters, tunnel colour scheme, weather, etc.

3 338 Ondrej Pribyl et al. / Procedia Engineering 192 ( 2017 ) The right side of the model indicates the impacts of the initial occurrence based on other factors that are usually related to the tunnel equipment, such as self-evacuation capability, ability to quickly identify the occurrence, ventilation efficiency, etc. The result is the frequency of the scenario occurrence given by the configuration of the marginal conditions. The impact of the occurrence expressed by the number of injured or killed people are determined using various estimation methods based mostly on exposure of a person or a group of people to toxic smoke. Fig. 1 Bowtie model The probabilistic methods are reliable only in case when they are based on representative samples of real statistical values. For example, it is necessary to have the statistics of fire occurrences during the peak traffic hours, malfunctions, wrong functionality of ventilation in case of fire and tens of other statistically relevant combinations. The Austrian regulation for calculation of risk using the probabilistic method [4] is based on statistics of 447 accidents ranging from 1992 to 2003 during which people were injured. The basic statistical values for Germany were obtained by analysing 979 accidents in 80 tunnels in Germany. In case such statistical data are missing, the use of the probabilistic method is unreliable and it is better to use the recommendation of the TP229-Z1 and apply the Scenario Analysis Method. To conclude, for the probabilistic methods in general it can be said that this kind of analysis is usually elaborated in most detail and therefore it requires the most time, is relatively expensive and mainly, it has to have enough input data, which is a requirement that in case of statistics of, for example, accidents and fires for different road tunnel types and marginal conditions is often not met Deterministic risk assessment overview and benefits The deterministic risk assessment approach can approximate a particular real life situation in a particular tunnel well. This is accomplished by including a knowledge base (scenario DB) in the software. The novelty of this approach is accomplished through a unique combination of different universal models incorporated into the software CAPITA. In the off-line mode, different scenarios for each relevant tunnel are simulated covering different traffic characteristics, different fire characteristics, but also different behaviour of people trapped in the tunnel. The behaviour is affected by the different technological safety systems in the tunnel, which makes the important link within the model. Another important feature of this approach is the simplicity of incorporating different parameters and implications of the new parameters are computed quickly. Contrary to the currently used probability models, the proposed solution is able to accommodate the dynamics of the processes. For example, the time needed for the people trapped in the tunnel to make a decision and start with evacuation can be easily incorporated. The result of the model, i.e. the number of endangered people and resulting estimate mortality rate, is computed in the following steps: The number of people trapped in different tunnel segments as determined by VISSIM. The behaviour and movement of people heading towards the emergency exits or tunnel portals is simulated using microscopic simulation tool EXODUS. This determines the time needed for clearing of the tunnel and the location of people in different time instants. The distribution of heat and smoke as a function of the tunnel characteristics, fire characteristics as well as the environmental characteristics is estimated using so called Computational Fluid Dynamics (CFD) software SMARTFIRE.

4 Ondrej Pribyl et al. / Procedia Engineering 192 ( 2017 ) The comparison of the fire and smoke propagation to the actual movement of people provides an estimate of the number of people endangered by the fire and the toxic smoke. The implicit model assumes that carbon monoxide and hydrochloric acid are the factors causing death during the fire. Their propagation in the height of 160 cm as provided by the software SMARTFIRE is used in the model. 3. Software CAPITA An overview of the models and their relations as defined in the software CAPITA is provided in Fig Model I: Traffic Model I provides the estimate of the number of people trapped in particular 50 m long tunnel segments. A microscopic simulation tool VISSIM from the company PTV is used to do this task. Since this is a stochastic simulation tool, the measurements have to be repeated for a particular scenario several times, and the average and standard deviation values are recorded for every 50 m long segment. There have been different scenarios taken into consideration with different parameters, such as different travel demand, different composition of the traffic flow, different driver behavior or for example different occupancy of vehicles. Just to give an example, the scenarios covered speed limits of 50 and 70 km/h for the urban road tunnels as well as 80 and 100 km/h for the rural road tunnels. The input traffic flow of 1,600, 800 a 160 vehicles per hour and the truck ratio of 0.10 and 0.30 was considered. The scenarios included also a bus, resulting in an increase of the people trapped in the tunnel. This model results in 32 different traffic scenarios with a recording of an average as well as maximum number of vehicles (of particular categories) for every 50 m long tunnel segment at a time Model II: Evacuation Model II estimates the number of people trapped in the particular tunnel segments. This is based on the output of Model I and an assumption about the average number of passengers in vehicles of particular type. The actual evacuation time is determined based on the simulation software EXODUS. Here, the different characteristics of the people (such as walking speed, speed of the decision making process and others) and a mix of genders and ages are combined to form additional different scenarios. The awareness and reaction time of trapped people is influenced by tunnel technological system which should be taken into consideration [5]. The mean walking speed was set to ost staveb Non- [6], [7] taking into account age distribution. The average walking speed is higher usi The results were captured in different scenarios showing the evacuation process in the tunnel Model III: Fire and smoke propagation Model III is using the simulation software SMARTFIRE to estimate the fire and smoke (carbon monoxide and hydrochloric acid in the default scenarios) propagation in the tunnel in case of fire. The different scenarios cover different intensity of fire, its location, or for example different environmental variables such as air velocity or the influence of the tunnel ventilation system. The software also allows to apply arbitrary physical model and set another threshold values. The time when the deadly concentration of the smoke reaches the exit is estimated for all available exits and tunnel portals Model IV: Analysis of consequences Model IV provides the estimate of the mortality rate. This is based on merging the Evacuation Model (Model II) and the Fire and Smoke Propagation Model (Model III). This clearly implicates how many people do not reach the exit in time before the smoke concentration exceeds the given threshold see Fig. 2.

5 340 Ondrej Pribyl et al. / Procedia Engineering 192 ( 2017 ) As the previous paragraphs have explained, the CAPITA software works with the predefined scenarios for the traffic model and for the evacuation behaviour of people. These can be prepared in advance and stored for the actual use in a knowledge base. The propagation of smoke and fire behaviour depends on the particular tunnel and the environmental conditions and shall be determined for each tunnel as well as fire type separately CAPITA Implementation example Fig. 2 Block scheme of the CAPITA model Software CAPITA is composed of several modules, but only two of them will be introduced in this paper since they represent the core of the generalized model. After the geometric parameters of the tunnel and traffic parameters are entered, the following output is generated using the stored data: total number of passenger and freight vehicles in the tunnel (average and variance ), number of passenger and freight vehicles at the first exit (average and variance ), bus presence in the first tunnel segment, total fill-up time of the particular tunnel segment. The evacuation process is vital for calculation of the mortality rate. CAPITA uses the standard population distribution: 50 % of males and 50 % of females with normal age distribution. The evacuation speeds used in the calculations are the following: of buildings Non-industrial buildings"), epending on age (person 20 years old 1.47 m/s; person 70 years old 0.7 m/s). The curves in Fig. 3 show the numbers of unique evacuees in time under the assumption that they are not affected by the propagating smoke. Vertical bars positioned on the top of each exits represent time needed for the critical concentration of toxic smoke to develop near the particular exit. Once that happens, evacuation using the affected exits is no longer possible. The screen of the last simulation step summarizes the results of the simulation total number of trapped people and vehicles, evacuation time and absolute and relative number of victims. It is possible to generate an XLS report containing complete simulation settings and results. From the report it is possible to get all relevant values related to the simulated scenario: length of the tunnel is 750 m, all of its 50 m segments were filled-,

6 Ondrej Pribyl et al. / Procedia Engineering 192 ( 2017 ) at the beginning of the evacuation each segment contained 53 people (top estimate), total number of trapped people was 265, fire of 50 MW intensity generated toxic smoke that reached critical concentration in the first segment in 320 s, in the second segment in 360 s, etc., 8 people did not manage to reach the exit before 320 s elapsed, 8 people died, i.e. mortality rate in this case is 3 % (out of 265 trapped people). We can say that is our simulation with the current parameters setting the mortality rate is too high and the existing safety measures are not sufficient. Fig. 3 Evacuation simulation for exits placed 150 m apart. The numbers represent the time (in seconds) needed for the critical concentration of toxic smoke to develop near the particular exit. Source: CAPITA software. 4. Conclusion In this paper, the authors presented a novel pragmatic approach to risk analysis in road tunnels. This framework and particular models allow decision makers to evaluate risks (expressed through the probability of mortality) in different road tunnels based on predefined scenarios. The CAPITA software can give basic overview about safety also for investors or decision makers who are not specialists in this field, which is a significant improvement of deployment risk analysis close to praxis. As a result, it can help to reduce the cost of safety systems in tunnel. CAPITA creates generic models for traffic and escaping of pedestrians accompanied by physical model describing fire behaviour and smoke propagation specifically to each tunnel. These models together can estimate the mortality rate based on different road tunnel characteristics and traffic, human and fire parameters. CAPITA fulfils the Directive 54/2004/ES requiring a risk assessment of a tunnel over its entire service life [8]. References [1] M. Towards complex system theory, Tutorial, In: Neural Network World 2015, vol.25, no.1, pp [2] B. Kohl et al., Current practice for risk evaluation for road tunnel users [3] TP 229 Safety in Road Tunnels Amendment No Z1), Ministry of Transport, November 2016, pp. 84. [4] - Schiene - Verkehr, Ausgabe 2006, pp. 39. [5] Effect of tunnel technological systems on evacuation time, Tunnelling and Underground Space Technology, 2014, vol. 44, p [6] E. Ronchi, C. Pasquale, J. Capote, D. Alvear, N. Berloco, A. Cuesta, The evaluation of different evacuation models for assessing road tunnel safety analysis, Tunnelling and Underground Space technology, 30 (2012), [7] ik, Strassen- und Eisenbahnbau, Nr. 90, 1992, pp [8] Directive 54/2004/ES: Directive of the European parliament and of the council on minimum safety requirements for tunnels in the Trans European road network.