Atmospheric Dispersion Modeling of Reacting Chemicals
Presented at the Conference on Computing in Environmental Resource Management,
held in Research Triangle Park, North Carolina, December 2-4, 1996
X. John Zhang, N. Albert Moussa, Daniel E. Groszmann, and Michael C. Masonjones,
BlazeTech Corporation 24 Thorndike Street, Cambridge, MA 02141
Allen B. Beach, Richard C. Crews, and Don D. Harrison,
Wright Lab Armament Directorate Eglin AFB, FL 32542-5434
ABSTRACT
An advanced source characterization and dispersion modeling system for instantaneously released reacting chemicals has been developed. The model considers the complex interaction of three important components: physical, chemical, and thermodynamic. The physical component includes dense cloud slumping or buoyant puff rise, cloud lift-off and passive dispersion. The chemical component includes reaction with ambient air with an entrainment limited rate; and the thermodynamic component includes evaporation and condensation. The model also includes several databases containing physical, thermochemical and toxic data for a large number of species.
Three cases are presented to illustrate the comprehensive capabilities of the model. First, an explosive release of a mixture of RDX, Viton A, Butyl Rubber and Paraffinic Oil is modeled. The explosion products continue to react with air in the high temperature fireball as well as in the cooler stem underneath. Second, a spill of saturated Boron Trichloride, which after evaporation can react with water moisture in the atmosphere, is modeled. In both of these cases, large amounts of reaction heat are released, and the cloud temperature, buoyancy and the resulting final rise are all increased. In the second case, the initially dense cloud may become lighter than that of air and the cloud may lift off from the ground and rise to a high altitude. Third, a parametric study is performed on the fireball size for the combustion of hydrocarbons with release amounts varying over several orders of magnitude. Available experimental results are used in validating the model output. The effect of exothermic heat release has the tendency to reduce the maximum ground concentration and dosage and thus is of particular interest to safety/environmental engineers and decision makers.
1. INTRODUCTION
Accidental releases or planned tests frequently involve materials that react with ambient air (Kao, 1994). Chemical species are transformed and large amounts of reaction heat are produced. For example, open burn and open detonation are used in demilitarization to destroy certain materials by burning or detonation. These activities involve both the initial explosive reactions as well as the reactions of explosion products with ambient air. Another example is the accidental release of industrial materials, such as Boron Trichloride (BCl3) or Uranium Hexafloride (UF6) (Hanna, Chang and Zhang, 1996). These materials, upon release, will evaporate or sublimate into their corresponding vapor phase and then react with water vapor in the air. The frequently encountered hydrocarbon fireball is also an example of the chemical reaction of released materials with oxygen in the air. Section 112(r) of Title III of the Clean Air Act Amendment (1990) requires a hazard assessment of the potential effects of accidental releases as well as a plan of action for an emergency response. Since most of the available air quality models do not deal with chemical reactions with large heat releases, the development and application of dispersion models for reacting chemicals is urgently needed.
Current treatments assume that all the reactions finish immediately after the release, e.g. AFTOX (Kunkel, 1991), etc. With an amount of air entrained based on a stoichiometric ratio, the temperature, reaction products and their amounts are calculated. Then, the puff rise and subsequent passive dispersion are obtained without involving further chemical reactions. While being simple and including some effects of reactions, these treatments have several shortcomings: (1) When complex reactions (multiple steps and multiple components) are involved, it is difficult to determine a priori the amount of air needed to complete reaction. (2) The reaction products and their amount are independent of the rate and history of the entrained ambient air. (3) Sometimes, the receptor is so close to the release location that the reacting puff passes by the receptor before the full completion of the reactions. The concentration modeled at the receptor may not be proper. (4) An accurate treatment of buoyant puff rise whose buoyancy is continuously fed by the reaction heat is ignored.
This paper provides a brief description of a newly developed model, ADORA, and some modeling results of its applications. ADORA (Atmospheric Dispersion Of Reacting Agents) is a new-generation model which includes exothermic/endothermic chemical kinetics, multi-phase and multi-component thermodynamic transformations, and buoyant/dense/neutral puff behavior for any instantaneous release. More than 100 common industrial chemicals are in the databases. The model is equipped with a Windows-based user interface.
2. PHENOMENOLOGICAL DESCRIPTION OF TWO RELEASE SCENARIOS
Release of materials that involve chemical reactions can be generally divided into two categories: explosive release and reacting spill. They are described below with an example of open burn and open detonation for the first category, and an accidental release of Boron Trichloride for the second category.
Demilitarization agencies frequently demolish unwanted materials with explosives. Shortly after the explosion, the pressure and temperature of the puff (which is composed mainly of explosive products) reach a maximum. Then the puff quickly expands and its pressure drops to ambient. A large amount of reaction heat is generated that drives the puff into the high atmosphere. Along its rising trajectory, the puff entrains ambient air and the reaction continues. Considerable transformation of chemical species and generation of reaction heat take place. As more air is entrained, the puff temperature is reduced, the rate of reaction is diminished as is the heat released. The puff eventually stabilizes at a certain height without rising further due to the stratification or the atmospheric turbulence. The quantities of interest are the products formed and their concentration distribution in both space and time. Most observations reveal that a fireball stem is attached to the rising fireball. Even though the species concentration in the stem may be low, it can be still of importance since it is closer to the ground and the potential damage effect to the environment can be significant.
Industrial accidental releases of chemicals such as Boron Trichloride may experience the following stages, see Figure 1 for a typical release. A fast flash is first encountered upon their release since most of these chemicals are stored as saturated liquids with pressures higher than the ambient. A certain amount of liquid is evaporated to vapor as the cloud pressure drops to that of the ambient pressure. As more agent vapor and water vapor become available due to evaporation and entrainment of the ambient air, an exothermic chemical reaction occurs that releases a large amount of heat and reduces the cloud density. The initially spreading dense cloud may become buoyant and lift-off from the ground. The final puff rise height is strongly affected by the reaction heat produced. After the puff reaches its final height, passive dispersion due to atmospheric turbulence dominates. It is clear that the puff lift-off may significantly reduce the toxic material concentration at ground.
The above two cases show that the identity of toxic chemical species and its maximum ground concentration are strongly related to chemical reactions. Therefore, proper inclusion of reactions is critical to the dispersion modeling of reacting agents.
3. COMPUTATIONAL IMPLEMENTATION
A complete modeling system for instantaneously released reacting chemicals with a graphical user interface has been developed. The interaction of three important components are included. They are:
(a) puff/cloud physics: this consists of buoyant puff rise, dense cloud spreading, lift-off, and passive dispersion. The released puff may experience alternating stages, from being initially denser than air to lighter than air, and display different transport and dispersion characteristics.
(b) chemical reactions: toxic species may be destroyed or formed along with the release of a large amount of reaction heat. The reaction rates are determined by the entrainment rate as well as the chemical kinetics.
(c) thermodynamic transformation: most chemical releases are in liquid or in solid phases. The time-dependent phase changes due to evaporation or sublimation are important.
A schematic diagram on the interaction is shown by a triangle in Figure 2. Three computational modules were developed for modeling each component. The dispersion module calculates the puff geometry and the air entrainment rate as the plume travels downwind. The chemistry module uses the air entrainment rate provided by the dispersion module to calculate the composition of the puff based on a reaction rate that is entrainment limited. The thermodynamics module calculates the puff temperature and density based on the new puff composition. The puff temperature and density then affect the dispersion module.
A computational flow chart is shown in Figure 3 for the interactions of these three modules. Within each time step, three substeps are used corresponding to the three modules. Since the available databases and formulation procedures for explosive release and reacting spill are different, we developed the three modules for each scenario as described in the following table:
Table 1. Three modules for modeling explosive release scenario and reacting spill scenario
|
In Table 1, PEP, Propellant Evaluation Program (Cruise, 1991), is a computer code for calculating high-temperature thermodynamic properties and performance characteristics of propellant systems. In the modeling of explosive releases, the chemical kinetic reaction rates are not available in most cases. We applied a quasi-equilibrium approximation using the thermochemical equilibrium code (PEP) to calculate the chemical compositions at each time step in a mixture with entrained materials. This approximation is valid when the mixture temperature is high and the reactions are fast.
It should be noted that as more and more cool air is entrained in, the mixture temperature drops, some of the product species may disappear unrealistically, and the equilibrium approximation becomes inaccurate. A rational is developed for allowing the amounts of these species to be frozen at certain specified temperatures before the equilibrium approximation breaks down. A table is provided for the most frequently encountered species that need to be frozen and the corresponding temperatures, based on some theoretical considerations with empirical validations.
Other effects, not specifically mentioned above, such as radiative heat loss, ground heat transfer, wind shear, particle deposition, averaging time and concentration fluctuations are also included.
4. MODEL STRUCTURE
Input Groups
Five input groups are used allowing users to input the various conditions.
(1) Chemical Species: three sets of chemical species and the related parameters can be specified which are: (a) released chemical species with the released amount, temperature and pressure; (b) the chemical species that need to be frozen as well as the associated freezing temperatures; and, (c) condensed chemical species that may deposit on the ground and the corresponding particle density and particle diameter.
(2) Location and Time: this group includes: current modeling time, release time, name of the release location, latitude and longitude, time difference between Greenwich mean time and local standard time, roughness length at the wind site and the release site, elevation (mean sea level) and the wind measurement height.
(3) Ambient Conditions: temperature, pressure, humidity, cloud cover and category, wind speed, direction (degree from North) and averaging time, and inversion layer height if it is available.
(4) Atmospheric Stability: four ways of specifying or computing: (a) specifying continuous stability parameter ranging from 0.5 to 6, (b) Pasquill-Gifford parameter ranging from A to F, (c) calculation based on the input of standard deviation of wind direction, and (d) calculation based on solar heating.
(5) Release Parameters: such as release height and velocity (horizontal and vertical).
Databases
Five databases are used in ADORA and users are allowed to add, edit, and delete data.
(1) Thermochemical Database, mostly reconstructed from PEP database (Cruise, 1991) which is based on JANAF (Chase, 1985), includes heat formation, enthalpy, entropy, and specific heat.
(2) Physical Data are obtained from various sources (Weast, 1968). These data include boiling point and saturation pressure.
(3) Reaction Equation (for the scenario of reacting spill only), has a list of stoichiometric equations for reactions with ambient air.
(4) Toxicity Data include: IDLH (Immediately Dangerous to Life and Health), TLV (Threshold Limit Value), and STIL (Short Term Inhalation Limits).
(5) Location. Several locations and the parameters associated with them are available, such as surface roughness and wind measurement height.
Running Modes
The user has the option to choose a single run for the inputted source condition and ambient conditions, or a sensitivity run by varying one or several parameters. The sensitivity run mode can be used to identify the worst case scenario, a frequently needed function in safety and environmental assessment. ADORA allows the user to arbitrarily vary four variables with any step size, they are: ambient stability, relative humidity, ambient temperature, and wind speed. The worst case can be defined by a user as having the maximum concentration of certain species at a certain height over all the possible cases.
Outputs
To fully delineate the dispersion modeling results, ADORA presents three types of output information for a user specified concentration averaging time.
(1) Cloud Summary: A table of results are included, i.e.: puff lift-off time and downwind distance, the location and puff condition at the transition to passive dispersion which includes the final cloud rise, the sizes of puff and tail (if it is modeled), and the downwind distance and time when the puff reaches the equilibrium height. A list of chemical species formed and the corresponding masses in the cloud as well as in the tail is provided at the equilibrium height. Based on these information, the passive dispersion of these species can be readily calculated.
(2) Concentration and Dosage: The concentration is averaged over a duration of interest by the user, either long or short. Dosage is the accumulation of concentration at any spatial point integrated over time. The local values of both concentration and dosage at any receptor location and time (user specified) are immediately available. Maximum values at any plane and any time are also calculated. Various concentration contours at any plane (horizontal as well as vertical) are available, these include the ground concentration contour of deposited material. The output also includes envelope contours, defined as the contour of maximum concentration or dosage at any location when the cloud passes through in the downwind direction, and toxic corridor, defined as the area outside which the concentration/dosage value is lower than a certain threshold for a specified confidence level.
(3) Downwind Evolution: The downwind characteristics of the puff are plotted as a function of downwind distance or time. The characteristics of interest include physical: puff size, temperature, density, and height, as well as chemical: the mass or concentration of each species in the puff.
These comprehensive outputs are useful in allowing the user in examining the concentration/dosage at interested location and time and conducting environmental and risk assessments.
5. SAMPLE MODELING RESULTS
ADORA is applied to the dispersion modeling of three cases. The release conditions and the corresponding results are summarized below.
Explosive Release
An explosion of RDX [C3N6H6O6] and Viton A [(C4H3F5)n], Butyl Rubber [(C4H8)n] and Paraffinic Oil [(C6H10)n] with total mass of 902.2 kg is released into the atmosphere. Summaries of input and output parameters are shown in Figure 4a. In the products, Nitrogen Monoxide (NO), Nitrogen Dioxide (NO2), Carbon Monoxide (CO) and Hydrogen Fluoride (HF) are frozen at temperature 1100 K, 1700 K, 2100 K, and 1100 K, respectively. The list of chemical species in the fireball as well as in the stem at the final rise height is shown in the output summary. The cloud reaches an equilibrium height of 490 m at downwind distance of 356 m after 1.7 minutes with a diameter of 208 m. Figure 4b presents the downwind evolution of concentrations of product species as a function of time before passive dispersion. It is seen that as the CO concentration is decreasing quickly in the first half second, NO and NO2 are formed. Beyond the first second, the concentration variation is dominated by the entrainment of ambient air into the cloud. Figure 4c shows the HF envelope concentration contour during passive dispersion in a vertical plane along the downwind direction. The tail effect is only significant close to the release point.
Reacting Spill
In this case, we consider the instantaneous release of 1000 kg of saturated liquid Boron Trichloride (BCl3) at a storage temperature of 302K. The liquid part of the chemical evaporates and the vapor part reacts with the moisture in the air. The ambient wind is 2.3 m/s measured at a 4.6 m height with a standard deviation in direction of 19 degrees. The ambient and ground temperatures are 302 K; relative humidity is 40%; roughness length is 3 cm. The modeling results are shown in Figure 5.
The two plots on the right show the evolution of the amount and concentration of BCl3 liquid, BCl3 vapor and reaction products HCl and H3BO3 versus downwind distance. The two plots on the left present the variations of the cloud temperature, density, volume and height as the cloud drifts in the downwind direction. Key transitions and characteristics of the cloud are described as follows.
The liquid agent is fully evaporated at 1.6 m. Before reaching that point, not much reaction has taken place. The puff lifts off the ground at 26 m due to the large amount of reaction heat released. Afterwards, the buoyant puff experiences a faster rise and the chemical reaction ends at 53 m. Then the cloud temperature drops quickly and the puff reaches equilibrium height at a downwind distance of 240 m. The ambient turbulence starts to dominate the puff motion and the dispersion afterwards. This is treated as passive dispersion and not shown in Figure 5. It should be pointed out that the cloud lift-off significantly reduces the maximum ground concentrations of toxic chemicals.
Hydrocarbon Fireballs
Figure 6 shows the modeling results of reacting hydrocarbon fireballs in the air (Moussa et al, 1995). Fireball diameter versus the combustion energy of the released chemicals are presented. Experimental data (open symbols) are presented for a number of fuels and propellants ranging from small to large-scale tests. The model results are plotted in solid symbols. Since the conditions for the measured fireball diameters were not clearly defined in these experiments, considerable scattering of the experimental data points is present. We calculated the fireball diameters corresponding to two times after release: (1) when the reacting puff reached its maximum temperature; and (2) when all fuels have been consumed. They are shown as solid circle and square symbols, respectively. The choice of these two times reflects part of the uncertainty in measuring the diameter. The model results fall on the mean of the measured test data with scattering around them. These results compare very well with both field and laboratory test results over seven orders of magnitude in energy release.
6. CONCLUSION
A source characterization and atmospheric dispersion model that predicts the product chemical species and their concentrations has been developed. The model includes the interactions of three interrelated modules: physical dispersion, chemical reaction, and thermodynamic transformations. Several databases are used to provide required data. Radiation, two-phase dynamics, exothermic and endothermic chemical reactions are included, the calculated reaction rates are dependent on the entrainment rates. ADORA is also equipped with a GUI (graphic user interface) and can be run on a PC 486+ machine with 16 MB RAM.
The model bridges the gap between source specification and meteorological dispersion by careful consideration of near-source effects including the product formation and exothermic heat release. The chemical transformation may convert toxins to non-toxins, or vice versa. The heat release increases the cloud buoyancy and the puff has a higher final rise and larger dilution rate. These will significantly reduce the maximum ground concentration and dosage. Thus, unnecessary conservatism in designing and planning for storage or test facilities can be relaxed.
This model has a wide range of applications such as: (a) reaction of halogenated compounds with moisture (UF6, SiCl4, TiCl4, BCl3, SF6, AsF3); (b) hydrocarbon combustion (C3H6, C6H6, C2H3Cl); (c) chemical releases triggered by explosives (open burn and open detonation, and the breakdown of HCFCs in fires: FM-200, etc.) The model can be used to meet the regulations requirements, e.g., Clean Air Act 112(r), Risk Management Plan (RMP), and PSM (29 CFR 1910), OSHA and DOT, for hazard/risk/safety assessment and preparing for Environmental Impact Statements (EIS).
To expand the potential application, we are interfacing ADORA with CHEMKIN (a general chemical kinetics program, Kee et al, 1989), GIS (Geographical Information System) and real time meteorological data input. The code may also be more applicable if complex terrain effects, building wake effects, and the inversion layer partition are included.
7. REFERENCES
Alan S. Kao (1994) Formation and removal reactions of hazardous air pollutants, J. Air & Waste Manage. Assoc. 44: 683-696.
Chase M.W.Jr., Davies C.A., Downey J.R.Jr., Frurip D.J., McDonald R.A., and Syverud A.N. (1985) JANAF Thermochemcial Tables, Third Edition, Journal of Physical and Chemical Reference Data, 14.
Cruise D. R. (1991) Theoretical Computations of Equilibrium Composition, Thermodynamic Properties, and Performance Characteristics of Propellant Systems. NWC TP 6037, Naval Weapons Center, China Lake, CA 93555-6001.
Hanna S.R., Chang J.C. and Zhang X.J. (1996) Modeling accidental releases to the atmosphere of a dense reactive chemical (Uranium Hexafluoride), Atmospheric Environment (To appear).
Kee R.J., Rupley F.M., and Miller J.A. (1989) Chemkin-II: A Fortran Chemical Kinetics Package for the Analysis of Gas-Phase Chemical Kinetics, Sandia Report, SAND89-8009, UC-401.
Kunkel, B. A. (1991) User's guide for Air Force Toxic Chemical Dispersion Model (AFTOX). AFGL-TR-88-0009, AFGL, Hanscom AFB, MA 01731, U.S.A.
Moussa N. A., Zhang X. J., Groszmann D. E., Beach A. B., and Noland R. B. (1995) Reacting Puff Rise, JANNAF Propulsion and Subcommittee Joint Meeting, 4-8 December, 1995. Tampa, Florida.
Weast R.C. (ed.) (1968) Handbook of Chemistry and Physics. The Chemical Rubber Co. 18901 Cranwood Parkway, Cleveland, OH 44128.