DK^SO Ris0-R-1O3O(EN) Dense Gas Dispersion in the Atmosphere Morten Nielsen Wti 3 0 1998 OSTl I y§MBUTm OF -nj c ■cl /'-r.'A J Ris0 National Laboratory, Roskiide, Denmark September 1998 DISCLAIMER Portions of this document may be illegible in electronic image products. Images are produced from the best available original document. Ris0-R-1O3O(EN) Dense Gas Dispersion in the Atmosphere Morten Nielsen Ris0 National Laboratory, Roskilde, Denmark September 1998 8 Abstract Dense gas dispersion is characterized by buoyancy induced gravity currents and reduction of the vertical mixing. Liquified gas releases from industrial accidents are cold because of the heat of evaporation which determines the den­ sity for a given concentration and physical properties. The temperature deficit is moderated by the heat flux from the ground, and this convection is an additional source of turbulence which affects the mixing. A simple model as the soil heat flux is used to estimate the ability of the ground to sustain the heat flux during release. The initial enthalpy, release rate, initial entrainment and momentum are discussed for generic source types and the interaction with obstacles is considered. In the MTH project BA experiments sourcee with and without momentum were applied. The continuously released propane gas passed a two-dimensional remov­ able obstacle perpendicular to the wind direction. Ground-level gas concentrations and vertical profiles of concentration, temperature, wind speed and turbulence were measured in front of and behind the obstacle. Ultrasonic anemometers pro­ viding fast velocity and concentration signals were mounted at three levels on the masts. The observed turbulence was influenced by the stability and the ini­ tial momentum of the jet releases. Additional information were taken from the ‘Dessert Tortoise ’ ammonia jet releases, from the ‘Fladis’ experiment with transi­ tion from dense to passive dispersion, and from the ‘Thorney Island’ continuous releases of isothermal freon mixtures. The heat flux was found to moderate the negative buoyancy in both the propane and ammonia experiments. The heat flux measurements are compared to an estimate by analogy with surface layer theory. The present report is a slightly revised version of a Ph.D. thesis originally sub­ mitted to the Technical University of Denmark (DTU) on December 31, 1997. Prof. Dr. FI. Bo Pedersen from the Department of Hydrodynamics and Water Re­ sources (ISVA) acted as supervisor together with Dr. N.O. Jensen from the Wind Energy and Atmospheric Physics Department (VEA), Ris0 National Laboratory. Dr. N.E. Ottesen Hansen and Prof. Dr. T. Fannelpp acted as external examiners. ISBN 87-550-2362-2 ISSN 0106-2840 Information Service Department • Risp • 1998 Contents 1 Introduction 5 2 Box models 6 3 Dense gas fronts 11 4 Mixing and entrainment 18 5 Obstacles 30 6 Concentration fluctuations 37 7 Surface temperature ^0 8 Dense gas sources 45 9 Cloud density 58 10 More on box models 71 11 Propane experiments with obstacles 82 12 Large-scale ammonia experiments 125 13 More ammonia experiments 135 14 Isothermal freon experiments 154 15 Conclusions 168 Acknowledgement 171 References 171 Summary 183 Sammendrag — summary in Danish 186 A The MTH Project BA propane experiments 191 B The FLADIS ammonia experiments 198 C Gas concentrations from sonic anemometers 219 D Anemometer coordinate transformations 226 E Thermo dynamic background 227 F Fluid mechanical background 236 Ris0-R-1O3O(EN) 3 G Micrometeorological background 246 H Literature guide 257 I Substance properties 234 Notation 260 Index 273 4 Ris0-R-1O3O(EN) 1 Introduction In recent decades it has been realized that gas clouds of negative buoyancy disperse in a way quite different from that of a passive tracer. The gravity force influences a dense gas cloud in two ways - it generates internal currents, and the stratification between the dense cloud and the lighter air above reduces the turbulent mixing. In this situation the theory of ordinary atmospheric dispersion is insufficient. A practical definition of the problem is: problem definition atmospheric dispersion affected by buoyancy forces acting on the cloud volume as a whole. This includes dispersion of substances with low molecular weight provided that the temperature is low enough to create a sufficient density surplus. On the other hand dispersion of dilute mixtures of gases with high molar weight - for instance a tracer gas like SFg - is considered to be an ordinary dispersion problem. Clouds, which are dense because of small particles, are not considered directly, but they should behave similarly if the deposition time is long compared to the time scale of the dispersion, see eg Bettis, Makhvildze & Nolan (1987) and Bonnecaze, Hallworth, Huppert & Lister (1995). The duration of a typical dense gas dispersion phase is only a few minutes - depending on the release size and the wind conditions. Dense gas clouds of concern emerge from accidental releases in industrial process applications units or during transport. This is not an every day event, but a risk of handling toxic or inflammable gases. Toxic gases is a hazard to the staff and in some cases even to a nearby population. Fire and explosions may cause severe physical damage and injuries, but the hazardous concentrations of flammable gases are usually much higher than those of toxic gases, so the endangered area is smaller. In practice, risk management considers a lot of topics which have no accurate answers. These matters could be: - recognition of relevant scenaria and probability estimates - the toxic effects on human beings and environment - the size of the endangered population - the risk of ignition and the physical damages after fire or explosion - possible mitigation methods - the benefits of improved training, regular inspection, and alarm systems. Dispersion modelling plays an important role in risk analysis, but it is not the only issue. The risk engineer and his client do usually not ask for fine details on the process. Most people are familiar with the theatrical effect, when a mist of smoke covers related phenomena the stage. This artificial cloud is often dense and flows along the floor, so it is a relevant picture for visualisation. Gravity currents are also important in geophys ­ ical phenomena like avalanches, volcanic emission, katabatic winds and maritime dense bottom currents. It is possible to draw parallels between these research areas and the fluid mechanical part of the dense gas problem. Organization of the present report As hinted by the title the objective is to learn from dense gas field experiments. This point-of-view has led to particular interest in the thermodynamics of cold gas clouds, source dynamics, turbulence characteristics and the effects of variable wind directions. The report is organized in the following main parts: Ris0-R-1O3O(EN) 5 Chapter 2—10: The physics of dense gas dispersion. Chapter 11-14: Studies of selected field data. Appendices: Material for selective reading. Acknowledgement , references and summaries in English and Danish are inserted before the appendices. List of notation and topic index are found at the end of the report. Dense gas experts will probably take most interest in the chapters which discuss experimental data. Chapter 2 gives a brief introduction to the widely used dense gas box models. dense gas physics This is followed by two chapters on dense gas dynamics, ie chapter 3 on dense gas fronts, chapter 4 on the mixing with ambient air. Chapters 5 and 6 discuss the effects of obstacles and the concentration fluctuations of a spreading gas cloud. Heat transfer from the ground to a cold gas cloud gives internal convection which affects the turbulence, and in chapter 7 the time development of this process is examined. Boundary conditions like source strength, initial momentum, liquid fraction and temperature are important for the dispersion, and chapter 8 gives an introduction to the physics of typical gas sources. Chapter 9 describes how to calculate the cloud density including the effect of a non-isothermal release. The box models introduced in chapter 2 are further discussed in chapter 10. In chapter 11 continuous release experiments with liquified propane are analysed. experiments The effects of initial momentum and an obstacle perpendicular to the wind direc­ tion are discussed with emphasis on the internal cloud turbulence. Chapter 12 contain a description of some large-scale ammonia dispersion experiments with focus on details complementary to the information from the propane experiments, in particular measurements of the heat flux from the ground. In chapter 13 an­ other set of ammonia experiments is used to demonstrate the effects of plume meandering. In chapter 14 turbulence data from an isothermal freon experiments without heat convection is discussed. Appendices A and B present details on the experiments analysed in chapters 11 appendices and 13. Appendix C describes a method for estimates of fast gas concentration time series based on sound virtual temperature measurements from sonic anemometers with attached thermocouples. Appendix D explains the geometrical transforma­ tion used for the alignment of three dimensional wind speed time series after the mean wind direction. Thermodynamic, fluid mechanical, and micro-meteorological concepts are presented in appendices E, F and G. Appendix H is a short introduc­ tion to experimental work, sources of literature and available data bases. Finally appendix I contains a collection of physical and toxicological gas properties. 2 Box models The interest in dense gas dispersion is stated by the potential hazards of inflam­ introduction mable or toxic chemicals. Risk assessment is usually based on complex informa­ tion from many disciplines, and it is often argued that a crude description of the dispersion problem is sufficient. Therefore dispersion models with fast solutions to ordinary differential equations and just a few model parameters have become popular, eg HEGADAS (Witlox 1994), DEG ADIS (Spicer & Havens 1986), SLAM (Ermak 1990), and DRIFT (Webber, Jones, Tickle & Wren 1992).