Thesis (M.S.) -- Worcester Polytechnic Institute. / Keywords: Hot smoke test; FDS; CFD; Computer modelling. Includes bibliographical references (p.105-110).

Understanding the burning behaviour of litter fuels is essential before developing a complete understanding of wildfire spread. The challenge of predicting the fire behaviour of such fuels arises from their porous nature and from the strong coupling of the physico-chemical complexities of the fuel with the surrounding environment, which controls the burning dynamics. In this work, a method is presented to accurately understand the processes which control the burning behaviour of a wildland fuel layer using numerical simulations coupled with laboratory experiments. Simulations are undertaken with ForestFireFOAM, a modification of FireFOAM that uses a Large Eddy Simulation solver to represent porous fuel by implementing a multiphase formulation to conservation equations (mass, momentum, and energy). This approach allows the fire- induced behaviour of a porous, reactive and radiative medium to be simulated. Conservation equations are solved in an averaged control volume at a scale su cient to contain both coexisting gas and solid phases, considering strong coupling between the phases. Processes such as drying, pyrolysis, and char combustion are described through temperature-dependent interaction between the solid and gas phases. Di↵erent sub-models for heat transfer, pyrolysis, gas combustion, and smouldering have been implemented and tested to allow better representation of these combustion processes. Numerical simulations are compared with experiments undertaken in a controlled environment using the FM Global Fire Propagation Apparatus. Pine needle beds of varying densities and surface to volume ratios were subject to radiative heat fluxes and flows to interrogate the ignition and combustion behaviour. After including modified descriptions of the heat transfer, degradation, and combustion models, it is shown that key flammability parameters of mass loss rates, heat release rates, gas emissions and temperature fields agree well with experimental observations. Using this approach, we are able to provide the appropriate modifications to represent the burning behaviour of complex wildland fuels in a range of conditions representative of real fires. It is anticipated that this framework will support larger-scale model development and optimisation of fire simulations of wildland fuels.

Det finns ett flertal brandsimuleringsprogram på marknaden som används i olika utsträckning varav ett är Fire Dynamics Simulator (FDS). En av nackdelarna med FDS är att det kan ta mycket tid att göra en brandsimulering. Det finns brandsimuleringsprogram som med stor sannolikhet utför brandsimuleringar snabbare än FDS. För några av dessa brandsimuleringsprogram finns det inte någon dokumentation om hur resultaten som brandsimuleringsprogrammen producerar ställer sig mot det som skulle hända i verkligheten vid en brand, något som kallas att vara valideratdet vill säga programmen är inte validerade. Ett av dessa brandsimuleringsprogram är Multi-Zone Fire Model (MZ-Fire Model). Brandsimuleringsprogrammet MZ-Fire Model bygger på ett multizonkoncept framtaget av Suzuki et al. Multizonkonceptet har validerats i tidigare studier varav ett är en brand i tunnel men även bränder i mindre lokaler har prövats. Det finns utrymme för ökad kunskap om hur multizonkonceptet hanterar bränder i stora rumslokaler då det inte finns någon känd dokumentation kring detta. Det finns i dagsläget inte en enda studie som behandlar brandsimuleringsprogrammet MZ-Fire Model. I rapporten redogörs för simulerande av en brand i 4 olika rum av brandsimuleringsprogrammen MZ-Fire Model och FDS, dess simulerade värden är sedan jämförda mot varandra. / There are a number of fire simulation programs on the market that are used to varying degrees, one of which is Fire Dynamics Simulator (FDS). One of the disadvantages of FDS is that it can take a lot of time to do a fire simulation. There are fire simulation programs that are very likely to perform fire simulations faster than FDS. For some of these fire simulation programs, there is no documentation on how the results produced by the programs compare with what would happen in the event of a real fire, something called to bethat is they are not validated. One of these fire simulation programs is Multi-Zone Fire Model (MZ-Fire Model). The fire simulation program MZ-Fire Model is based on a multi-zone concept developed by Suzuki et al. The multi-zone concept has been validated in previous studies, one of which is a fire in a tunnel but fires in smaller premises have also been tested. There is room for increased knowledge about how the multi-zone concept handles fires in large rooms, as there is no known documentation on this. Currently, there is not a single study dealing with the MZ-Fire Model program. The report describes the simulation of a fire in 4 different rooms by the programs MZ-Fire Model and FDS, its simulated values ​​are then compared against each other.

FDS (Fire Dynamics Simulator) använder en version av Navier-Stokes ekvationerna för att göra noggranna beräkningar av värme- och gastransport genom brandbelastade utrymmen. Utvecklarna av programmet arbetar kontinuerligt med att validera det allteftersom nya funktioner tillförs för att öka noggrannheten och bredda tillämpningsområdena. Väldigt lite av detta arbete fokuserar dock på FDS:s konduktionsmodell, den endimensionella Crank-Nicolson metoden. Det här examensarbetet ämnar därför undersöka noggrannheten i FDS:s konduktionsmodell genom jämförelse mot beräkningar med FEM (Finita elementmetoden). En FDS-modell skapades för att tillåta undersökning av en vägg och dess randvillkor med så liten påverkan från andra faktorer som möjligt. Detta för att skapa en kontrollerad omgivning som enkelt kunde replikeras i efterföljande FE-beräkningar av det konduktiva värmeflödet genom den solida obstruktionen. Tre väggar (10 cm betong, 20 cm betong och 1 mm stål) vardera med tre randvillkor (Exposed, Void och Insulated) utsattes för tre temperaturer (100 °C, 500 °C och 1000 °C) vilket ger 27 FDS simuleringar. Den adiabatiska yttemperaturen mättes i varje simulering och användes som indata till motsvarande FE-beräkningar. Resultatet påvisade inga signifikanta motsägelser vad gäller randvillkoren, med tillräcklig tid för termisk penetrering påverkade de den resulterande temperaturen som väntat. Undantaget var en mindre avvikelse i stålväggarna som utsattes för 100 °C och 500 °C med randvillkoren Exposed och Void där FDS aningen underskattade temperaturen relativt FE-beräkningarna. Gastemperaturerna i gridcellerna närmast väggen visade sig vara opålitliga. De tenderade att genomgå substantiella fluktuationer, troligen som ett resultat av hur FDS hanterar diskretiseringen av icke-solida volymer för Navier-Stokes beräkningarna. Dessa fluktuationer påverkade dock inte de resulterande solida temperaturerna eftersom medelgastemperaturen var korrekt. FDS påvisades även ha en tendens att aningen överskatta yttemperaturen under de första minuterna av simuleringarna relativt FE-beräkningarna. Temperaturerna från de två beräkningsmetoderna konvergerade dock efter några få minuter i samtliga tester. Dessa avvikelser ansågs ha för liten påverkan på de solida temperaturerna för att påvisa onoggrannhet i FDS. Därmed drogs slutsatsen att FDS:s beräkningar av temperaturer i solida material är tillräckligt noggranna inom dessa avgränsningar. / FDS (Fire Dynamics Simulator) uses a version of the Navier-Stokes equations to make accurate calculations of heat and gas flow through enclosures exposed to fire. The developers of FDS have, and continue to, validate it as new features get added in an attempt to increase its accuracy and broaden its potential applications. However, little of this effort is focused on FDS’ conductive heat transfer model, based on the one-dimensional Crank-Nicolson method. Thus, this study aims to test the accuracy of FDS’ conduction model by comparing it to calculations using FEM (Finite Element Method). FDS simulations were created so as to facilitate the study of a wall and its boundary conditions with as little interference from other factors as possible. This to create a controlled environment which easily could be replicated in the subsequent FE-calculations of the conductive heat flow through the solid obstructions. Three different walls (10 cm concrete, 20 cm concrete and 1 mm steel), each with the three different boundary conditions for the rear surface (Exposed, Void and Insulated), were exposed to three different temperatures (100 °C, 500 °C and 1000 °C) for a total of 27 FDS simulations. The adiabatic surface temperature was measured in each simulation in FDS and used as input for the corresponding FE-calculations. The results showed no clear inconsistencies in the boundary conditions, given enough time for thermal penetration they affected the resulting temperatures as expected. Save a slight deviation in the steel walls exposed to 100 °C and 500 °C with boundary conditions Exposed and Void where FDS slightly underestimated the temperature relative to the FE-calculations. The gas temperatures in the grid cells closest to the wall were found to be unreliable as they tended to undergo substantial fluctuations, likely as a result of how FDS handles the discretization of non-solid space for the Navier-Stokes calculations. These fluctuations were however not found to affect the solid temperatures as the mean gas temperature was accurate. FDS was also found to have a tendency to slightly overestimate the surface temperature in the first few minutes of the simulations relative to the FE-calculations. Though the resulting temperatures from the two methods converged after a few minutes at most in all tests. These deviations were considered to have too minor an impact on the solid temperature to justify claims of inaccuracy in FDS. Thus, the general conclusion of this study is that FDS’ predictions of solid phase temperatures are sufficiently accurate within these delimitations.

This research investigates the decomposition and burning behaviours of polyurethane foams experimentally and compares the experimental results obtained with the numerical results from the pyrolysis model of Fire Dynamics Simulator, Version 5 (FDS 5). Based on the comparison of model and experimental heat release rates, the accuracy of the pyrolysis model is quantified. In total, this research tested seven polyurethane foams consisting of three non-fire retardant (NFR) and four fire retardant (FR) foams. According to the simultaneous differential scanning calorimetry and thermogravimetric analysis (SDT) experiments, the decomposition behaviour of polyurethane foams under nitrogen environment is represented by two pyrolysis reactions. The first reaction consists of foam decomposition into melts and gases while the second reaction consists of the decomposition of the remaining melts into gases. The kinetic properties which govern the rate of decomposition are the activation energy (E), pre-exponential factor (A), reaction order (n) and heat of reaction (Δhr). Using graphical techniques, E, A and n of the first and second reactions are determined from the thermogravimetric analysis (TGA) results. Through analysing the differential scanning calorimetry (DSC) results, Δhr is determined from the changes in heat flow and sample mass. The thermophysical properties govern the heat transfer through material and these are the thermal conductivity (λ) and specific heat (cp) which are measured experimentally at ambient temperature through the Hot Disk method. Through the Sample Feeding Vertical Cone, the decomposition and melting behaviours of polyurethane foams in a vertical orientation are investigated and the foams tested can be categorised into those which produce melts only after ignition and those which produce melts and char after ignition. The 1-dimensional burning behaviour of foams is obtained from the cone calorimeter experiments. The NFR foams show a change from plateau burning behaviour at low heat flux to two stage burning behaviour at high heat flux while the FR foams consistently show two stage burning behaviour. The combustion property governs the amount of heat released when fuel combusts and this is the effective heat of combustion (Δhc,eff) which is determined from the heat released and mass consumed in the cone experiment. The 1-dimensional burning behaviour is simulated using the pyrolysis model of FDS 5 and two different modelling approaches are considered. The direct method uses the material properties determined experimentally as FDS 5 inputs while the refined method uses the genetic algorithm of Gpyro to refine the kinetic properties which are later used as FDS 5 inputs. The heat release rate of the model and experiment are compared through linear regression analysis which quantifies the accuracy of both methods. The accuracy is defined as the percentage of data points within the boundary of acceptance which is bounded by 25 % of the greatest experimental heat release rate. This assessment method places greater emphasis on the accuracy of developed burning phases and lesser emphasis on the accuracy of initial growth and final decay. The accuracy of the direct method is found to be 56 % while the refined method with estimated kinetic properties achieves a higher accuracy of 75 %. The 2-dimensional burning behaviours are investigated in the foam slab experiments for two different slab thicknesses, 120 and 100 mm. The opposed-flow spread of 120 mm slab is more intense and rapid while for the 100 mm slab, the flame spread is less intense and slow. FDS 5 is used to simulate the experimental results but when the material properties either developed experimentally or refined by Gpyro are used as inputs, the model fails to produce flame spread. This is because FDS 5 does not yet have the features which address the dynamics of foam melting and the reactive nature of the flame. In order to produce flame spread in the model, E of the reactions have been reduced to increase the decomposition rate.

"Fire investigative methodologies were researched and analyzed resulting in the development of an organizational tool to be used for conducting fire investigations. The tool, or field-guide, was designed to aid the investigator in processing structural fire scenes. The tool accomplishes this by providing, 1) thirteen forms for properly documenting the scene, 2) flowcharts which can enhance the investigators intuition for the fire’s growth rate and spread, and 3) basic engineering correlations which can be used to help validate hypotheses the investigator may develop. By employing these methods, the field-guide can be effectively used to lead an investigator through the entire investigation process - from data collection, to the formulation of hypotheses, and ultimately, to quantitative validation."

No description available.

No description available.

Proper Orthogonal Decomposition (POD), combined with the Method of Snapshots and Galerkin projection, is a popular method for the model order reduction of nonlinear PDEs. The POD requires the left singular vectors from the singular value decomposition (SVD) of an n-by-m "snapshot matrix" S, each column of which represents the computed state of the system at a given time. However, the direct computation of this decomposition can be computationally expensive, particularly for snapshot matrices that are too large to fit in memory. Hierarchical Approximate POD (HAPOD) (Himpe 2016) is a recent method for the approximate truncated SVD that requires only a single pass over S, is easily parallelizable, and can be computationally cheaper than direct SVD, all while guaranteeing the requested accuracy for the resulting basis. This method processes the columns of S in blocks based on a predefined rooted tree of processors, concatenating the outputs from each stage to form the inputs for the next. However, depending on the selected parameter values and the properties of S, the performance of HAPOD may be no better than that of direct SVD. In this work, we numerically explore the parameter values and snapshot matrix properties for which HAPOD is computationally advantageous over the full SVD and compare its performance to that of a parallelized incremental SVD method (Brand 2002, Brand 2003, and Arrighi2015). In particular, in addition to the two major processor tree structures detailed in the initial publication of HAPOD (Himpe2016), we explore the viability of a new structure designed with an MPI implementation in mind. / Master of Science / Singular Value Decomposition (SVD) provides a way to represent numeric data that breaks the data up into its most important components, as well as measuring how significant each part is. This decomposition is widely used to assist in finding patterns in data and making decisions accordingly, or to obtain simple, yet accurate, representations of complex physical processes. Examples of useful data to decompose include the velocity of water flowing past an obstacle in a river, a large collection of images, or user ratings for a large number of movies. However, computing the SVD directly can be computationally expensive, and usually requires repeated access to the entire dataset. As these data sets can be very large, up to hundreds of gigabytes or even several terabytes, storing all of the data in memory at once may be infeasible. Thus, repeated access to the entire dataset requires that the files be read repeatedly from the hard disk, which can make the required computations exceptionally slow. Fortunately, for many applications, only the most important parts of the data are needed, and the rest can be discarded. As a result, several methods have surfaced that can pick out the most important parts of the data while accessing the original data only once, piece by piece, and can be much faster than computing the SVD directly. In addition, the recent bottleneck in individual computer processor speeds has motivated a need for methods that can efficiently run on a large number of processors in parallel. Hierarchical Approximate POD (HAPOD) [1] is a recently-developed method that can efficiently pick out the most important parts of the data while only accessing the original data once, and which is very easy to run in parallel. However, depending on a user-defined algorithm parameter (weight), HAPOD may return more information than is needed to satisfy the requested accuracy, which determines how much data can be discarded. It turns out that the input weights that result in less extra data also result in slower computations and the eventual need for more data to be stored in memory at once. This thesis explores how to choose this input weight to best balance the amount of extra information used with the speed of the method, and also explores how the properties of the data, such as the size of the data or the distribution of levels of significance of each part, impact the effectiveness of HAPOD.

Die Digitalisierung im Bauwesen schreitet immer weiter voran und während in diesem Zusammenhang oftmals das Stichwort Building Information Modeling (BIM) fällt, entwickeln sich Disziplinen wie das Brandschutzingenieurwesen (BSI) unabhängig weiter. Das Brandschutzbüro Brandschutz Consult Ingenieurgesellschaft mbH Leipzig (BCL) verwendet das BSI, um ingenieurtechnische Verfahren heranzuziehen. BCL verfolgt als Unternehmensphilosophie das Ziel, mit neuen Methoden und Erkenntnissen ständig die eigenen Prozesse zu optimieren und zu erweitern. Unter diesem Gesichtspunkt soll in dieser Arbeit in Kooperation mit BCL untersucht werden, inwieweit sich die Ergebnisse aus einer Brandsimulation, darunter besonders der Rauch, in einer virtuellen Realität (engl. Virtual Reality (VR)) darstellen und in bestehende oder potenzielle Anwendungsfälle integrieren lassen. Dazu soll zunächst mit einer Betrachtung der brandschutztechnischen Grundlagen inklusive des BSIs und einer Analyse zum Stand des Brandschutzes in BIM begonnen werden. Im nächsten Schritt sind für die Brandsimulation bestimmte Fragen zu klären, wie z.B. eine entsprechende Berechnung technisch abläuft und welche Ausgabedaten und -formate eine solche Simulation bereitstellt. Zur Darstellung der Simulationsergebnisse in virtuellen Realitäten werden Grafik.Engines benötigt, die VR-Anwendungen ermöglichen. Wichtige Untersuchungsgegenstände sind z.B. die anwendbaren Programmier- und Skriptsprachen, mit deren Einsatz die Daten eingelesen und visualisiert werden können. Für die gefundenen Grafik-Engines wird dann recherchiert, ob es bereits bestehende Anwendungen oder Prozesse zur Darstellung von Brandsimulationen gibt. Ist dies der Fall, sollen deren Workflows untersucht werden, um anschließend ihre grundsätzliche Einsatzfähigkeit zu bewerten und Verbesserungsvorschläge zu äußern...:1. Prozesse im Brandschutz 1.1. Brandschutztechnische Grundlagen 1.2. Angewandte Ingenieurmethoden 1.3. Brandschutz mit Building Information Modeling 2. Ablauf einer Brandsimulation 2.1. Verfügbare Software 2.2. Aufbau einer FDS-Eingabedatei 2.3. Generieren von Simulationsdaten in FDS 2.4. Ausgabedaten und -formate 3. Software zur Darstellung in VR 3.1. Blender 3.2. Unity Engine 3.3. Unreal Engine 3.4. Vergleich der Engines 4. Visualisierung der Brandsimulation 4.1. Konzept der Datenübertragung 4.2. Bestehende Workflows für VR-Programme 4.3. Versuchsdurchführung 4.4. Auswertung der Versuche 5. Anwendungsfälle und Optimierungspotenzial 5.1. Potenzielle Einsatzmöglichkeiten 5.2. Optimierungspotenzial 6. Fazit A. Beispielmodell Blender B. Beispielmodell VRSmokeVis C. Prüfmodell Abkürzungsverzeichnis Abbildungsverzeichnis Tabellenverzeichnis Literaturverzeichnis / Digitization in the construction industry is progressing and while the keyword Building Information Modeling (BIM) is frequently mentioned, disciplines like the fire safety engineering are also evolving independently. The fire protection office Brandschutz Consult Ingenieurgesellschaft mbH Leipzig (BCL)) uses fire safety engineering for including engineering procedures. As a corporate philosophy BCL pursues the goal of constantly optimizing and expanding its own processes with new methods and scientific findings. From this point of view, in cooperation with BCL, this master thesis will examine to which extent it is possible to visualize the results of a fire simulation, in particular including the smoke, in Virtual Reality (VR) and to integrate them into existing or evolving applications. For this purpose, a consideration of the fire protection basics including fire protection engineering and an analysis of the status of fire protection in BIM has been started. In the next step the fire simulation must be investigated, i.e. how the corresponding calculation technically works and which output data and formats such a simulation provides. Graphic engines that enable VR applications are required to display the simulation results in VR. Important objects of investigation are e.g. the applicable programming and scripting languages. Those scripting languages are used to import and visualize the data. For the graphic engines found, research is initiated to determine whether there are already existing applications or processes for displaying fire simulations. If this is the case these workflows should be examined in order to subsequently evaluate their fundamental usability and to express suggestions for improvement. If possible, some of the optimizations should be carried out. Based on the existing processes in fire protection helpful application options are derived, for which the use must be proven in future projects.:1. Prozesse im Brandschutz 1.1. Brandschutztechnische Grundlagen 1.2. Angewandte Ingenieurmethoden 1.3. Brandschutz mit Building Information Modeling 2. Ablauf einer Brandsimulation 2.1. Verfügbare Software 2.2. Aufbau einer FDS-Eingabedatei 2.3. Generieren von Simulationsdaten in FDS 2.4. Ausgabedaten und -formate 3. Software zur Darstellung in VR 3.1. Blender 3.2. Unity Engine 3.3. Unreal Engine 3.4. Vergleich der Engines 4. Visualisierung der Brandsimulation 4.1. Konzept der Datenübertragung 4.2. Bestehende Workflows für VR-Programme 4.3. Versuchsdurchführung 4.4. Auswertung der Versuche 5. Anwendungsfälle und Optimierungspotenzial 5.1. Potenzielle Einsatzmöglichkeiten 5.2. Optimierungspotenzial 6. Fazit A. Beispielmodell Blender B. Beispielmodell VRSmokeVis C. Prüfmodell Abkürzungsverzeichnis Abbildungsverzeichnis Tabellenverzeichnis Literaturverzeichnis