Releases of toxic substances to the atmosphere are an unfortunate reality of an industrial society. The discharge of toxic material may occur either as a routine release or an accidental release. The consequences of toxic chemical releases can range from a simple nuisance to a virtual destruction of land, bodies of water or human injury and death.

In the United States, toxic releases from industrial sources and accidents have been addressed through various regulations, i.e., Superfund Title III: State Community Right-To-Know and Emergency Planning Laws, and the U.S. Department of Transportation's Hazardous Material Transportation Act. Before 1990 many state agencies developed their own regulations for toxic releases because the federal regulations were not clear on air toxins. However, the Clean Air Act Amendments of 1990 clearly specify the need for the determination of the impact of air toxins.

As a result of new regulations, more than 66,000 chemical process industries across the United States are required to prepare Risk Management Plans (RMP) under 1990 Clean Air Act's Title III, Section 112(r) if they store or use certain hazardous chemicals. Aforementioned rule covers 77 toxic chemicals and 63 flammable chemicals if they are used or stored in amounts in excess of regulated limits. Toxic chemicals include gases,
liquefied gases, liquids and solutions. Flammable chemicals include both gases and liquids.

Many companies have already prepared their plans while others are at different stages of preparation. This section provides helpful hints for the preparation of an RMP (including modeling) for those who have not completed their plans. The industry is required to submit their RMP by June 21, 1999. Noncompliance with the program is expensive and fines can be $10,000 to $25,000 per day, with up to a year in jail for each day of noncompliance or falsification.


The final RMP promulgation (U.S. EPA, 1996) is in accordance with the rule, which requires the development of a risk management program for industries that store or use chemicals on the list that includes the following elements:

Any facility that is required to submit an annual Toxics Release Inventory (TRI) Form R to the U.S. Environmental Protection Agency (EPA), or is subject to the U.S. Occupational Safety and Health Administration's (OSHA) Process Safety Management regulations, is likely to be regulated under the RMP rule. However, the amount of specific chemicals that could be involved in a possible catastrophe at the facility will determine the applicability of RMP rule. A threshold quantity (TQ) is specified for each regulated substance. The TQ varies between 500 lbs to 20,000 lbs for toxic chemicals; the TQ is 10000 lbs for all flammables. Detailed information is available on RMP on your web browser at

Facilities subject to their regulations must assess all potential hazards and failure modes, analyze worst-case scenarios of potential chemical release, determine one other most probable release, perform off-site consequence analysis, prepare emergency-response plans, and coordinate them with the Local Emergency Response Planning Committees.

The facilities subject to these regulations are divided into three program areas under the RMP rule:

Program 1 requires all facilities to evaluate the worst-case scenario, and one other most-probable incident for each regulated chemical stored or used onsite. The risk management plan must include the facility's five-year accident history and must demonstrate formalized coordination with any projected emergency response with the Local Emergency Planning Committee.

In addition to Program 1 requirements, Program 2 calls for a hazard assessment. The requirements for this assessment are not as comprehensive as Program 3 requirements.

Program 3 additionally requires a management structure to oversee the development, implementation and integration of the risk management program elements and a hazard assessment.

An organized approach is required for the development of the RMP. Typical steps are:


The key element of a plan is to determine the impact of a release. Atmospheric dispersion modeling of chemical releases, vapor cloud fires and explosions are important components in evaluating risks from released hazardous chemicals during the preparation of an RMP. Hanna and Drivas (1996) prepared guidelines for use of vapor cloud dispersion models in 1987 and revised it in 1996. The American Institute of Chemical Engineers prepared a workbook of test cases for vapor cloud source dispersion models (Hanna and Strimaitis, 1989).

The Environmental Protection Agency prepared a workbook of screening techniques for assessing impacts of 18 different types of air toxic release scenarios (U.S. EPA, 1988). The EPA released a computer program TSCREEN, based on the above workbook, in 1990. EPA revised the workbook in 1992 (U.S. EPA, 1992a). Kumar and Rao (1990) give a list of the programs available for managing chemical hazards. Latest guidance on the application of refined models for air toxic releases was published in 1993 (U.S. EPA, 1993). A summary of the guidance on dispersion models for toxic releases is given in Table 1. A comprehensive discussion on vapor cloud fires and explosion is given in "Guidelines for Evaluating the Characteristics of Vapor Cloud Explosions, Flash Fires, and BLEVEs" (AIChE, 1994).

Table 1

Guidance Models for Air Toxic Releases

Type of Modeling Models
Screening Technique

Refined Technique - Non-dense Gas

Refined Technique - Dense Gas


EPA Air Quality (U.S. EPA, 1986)

Guideline Models 


During the last 15 years, consulting companies, research organizations and other government agencies have developed a number of computer programs. A listing of all the programs currently available from EPA is available on the site at

Preparation of an RMP requires information on emission rates for worst-case scenarios and one other most probable release. The computer program ERate 1.1.1 described in the book by Kumar and Vashisth (1999) will help you determine release rates for different accident scenarios. Once the information on release rates is available, you will need an air quality model or "look-up tables" provided by EPA to perform offsite consequence analysis. Look-up tables are useful for companies without air modeling capabilities. The results for air contaminant concentrations obtained using these tables are conservative and should be used with caution.


In order to assist the reader in selecting a model for atmospheric dispersion, a brief description on five screening models: TSCREEN, ARCHIE, ALOHA/CAMEO, DEGADIS, and RMP*Comp which are widely used in the United States, is included. The choice of these models in this discussion is not an indication that other models can not be used. The selection and application of a model really depends on the release scenario. The sources and the cost of obtaining each model are given in Table 2.

Table 2

Source and Cost of the Screening Models

Model Source Telephone No./ e-mail Cost

CAMEO suite




FEMA, Publications Office
500 C Street, S.W.

Washington, DC 20472

National Safety Council

444 N. Michigan Ave.
Chicago, IL 60611
Same as CAMEO suite

Office of Air Quality
Planning & Standards Research
Triangle Park, NC 27711
Office of Response and Restoration

(312) 972-3275

Ext. 1300

(919) 493-3536

Free (order through the local FEMA office)

$1,500 (approx.)

$610 (approx.)

Available from website

Available from website

Each model is briefly described below.

    TSCREEN Model

The model is an easy to use computer program. It was developed by Pacific Environmental Services, Inc. to model emissions of toxic chemicals and their subsequent dispersion (concentration as a function of distance) from one of many different types of possible releases. The EPA models SCREEN, RVD and PUFF are grouped in the TSCREEN model and they are used to model the various scenarios according to the situation.

TSCREEN is programmed to automatically:

TSCREEN is an IBM PC-based software application written and compiled in Microsoft Basic Version 7.0 and Microsoft C version 5.1. The program requires 500 kilobytes of free random access memory (RAM). The TSCREEN program files occupy about 1 MB of disk space. The hard disk space needed will increase as scenario files are created. To install TSCREEN on hard disk systems make sure there is at least 1 MB of free disk space available to load and execute the program. A well-written user's guide is available from the EPA (U.S. EPA, 1994).

A new version of TSCREEN can be down loaded from the EPA's Internet site (www.epa.ttn/scram) and includes several changes to the original program.

    ARCHIE Model

The Automated Resource for Chemical Hazard Incident Evaluation (ARCHIE) model is designed for evaluating potential consequences of hazardous materials accidents by local emergency planning committees and industry. The program was developed by U.S. Department of Transportation in association with Federal Emergency Management Association (FEMA) and U.S. EPA. The program requires detailed input such as quantities of chemical released and physical conditions of releases. The ARCHIE program should be used in conjunction with the brown book (Anonymous, 1989). ARCHIE can be used to develop a variety of scenarios for incidents involving toxic, flammable and explosive hazards. The model uses a Gaussian equation for air dispersion calculation.

The program uses 717 Kb of disk space. It takes about 10-15 minutes to run the program including time required to respond to all the questions for a typical problem on an IBM compatible personal computer. The model is very user friendly.


The National Oceanic and Atmospheric Administration (NOAA) and the EPA developed the current version of the program with distribution and technical support provided by the National Safety Council. The Aerial Locations of Hazardous Atmospheres (ALOHA) model is the light-gas dispersion model and DEGADIS is the heavy gas model used in conjunction with CAMEO. ALOHA and DEGADIS estimate airborne pollutant concentrations downwind from a spill. The complete CAMEO system is designed for emergency planning and life-saving responses to chemical accidents.

ALOHA package is available in Apple Macintosh and IBM PC versions. ALOHA consists of multiple, integrated databases and programs and is completely menu driven. The chemical database, the mapping module and other emergency planning tools are accessible from the main menu screen. The latest version of ALOHA permits sophisticated modeling by estimating the area of the cloud and concentration overtime, given various environmental conditions. ALOHA's new modeling capabilities include: complex source routines for tanks, puddles, pipes, gas dispersion and indoor air filtration calculations. Therefore, the model could be used for refined modeling. A user's guide is available and is quite explanatory.

ALOHA runs on Apple Macintosh with five megabyte of random access memory (RAM) and 20 MB free hard disk space. It will run much faster on a Macintosh with a math co-processor chip. It requires at least two megabytes of hard disk space available to load ALOHA. ALOHA will also work on computers that do not have the math co-processor chip.

ALOHA will run on PC compatibles with Windows 3.1 or Windows 95. A minimum of 8 MB RAM and 30 MB free hard disk space is required.

    DEGADIS Model

The DEGADIS (Dense Gas Dispersion) was developed for the U.S. Coast Guard and the Gas Research Institute. It simulates the three phases of low initial momentum release of heavy gases: (1) the negative buoyancy-dominated dispersion phase, (2) the stratified shear flow phase and (3) the passive dispersion phase. The model is publicly available from EPA's web site. More user-friendly versions of the model can be purchased from consulting companies.


The RMP*Comp program performs offsite consequence analysis recommended by EPA under the RMP rule. This program was developed by NOAA and EPA. The 1999 version of the program handles vapor cloud fires for flammable gases liquefied under pressure. The generic distance-tables from Offsite Consequence Analysis Guidance are used for most chemicals to determine toxic endpoints. Separate tables are given for ammonia, chlorine and sulfur dioxide. No interpolation is performed from the distance-tables. Instructions for downloading, installing, and running the program are available from NOAA web site.


Atmospheric dispersion modeling discussed above is used to determine the distance to the endpoint for each toxic chemical released from a source. Endpoint concentrations are given in Appendix A to 40 CFR 68. These concentrations are less than the corresponding health indicators given by OSHA, NIOSH or ACGIH.

The endpoint for flammable compounds is defined as the point where the pressure will be equal to 1 psi as a result of explosion. Minor building damages are a result of this level of pressure.


The best way to use TSCREEN, ALOHA and ARCHIE models is to utilize the following interactive menu options in the order given below:

The screens are user friendly and let you walk through various phases of the models. You should not encounter any major difficulties during the use of these models

DEGADIS can be run on a personal computer and on a DEC VAX computer. The model is executed via batch files (with extension BAT) under DOS or via command files (with extension COM) under VMS. There are three programs to assist the user in preparation of input files and batch/command files.

RMP*Comp is an Oracle Media Objects "stack" that runs in Microsoft Windows 3.1, 95, 98, and NT and on Macintosh computers. The program is easy to use after installation.


Any discussion on models is not complete without some remarks on the ability of the models to predict reality. Air quality modeling became an essential tool for determining the compliance with the regulations. Therefore, it is important that the predictions made by an air quality model are reliable. Two types of performance measures are used to evaluate air quality models:

Difference measures represent a quantitative estimate of the size of the differences between observed and predicted values. Correlation is the quantitative measure of the association between observed and predicted values.

A model's ability to predict air pollution levels under changing conditions can only be tested after field measurements are taken under similarly changing conditions. The foregoing requirements cause the calibration of models to be a very expensive and often time-consuming study. The usual way to evaluate the predictions from a model is to draw a scatter diagram using predicted values and observed values. A variation of this approach is by computing the ratio of the predicted to the observed value. Old literature in the fields of science and engineering is full of such examples. Typically the ratio of observed value to predicted value of a good model, should not exhibit any trend with variables such as wind speed and stability class, and should not exhibit large deviations from unity (implying a perfect match between the model and observed). Later on correlation coefficient between the observed and predicted values became a popular way of looking at the performance of a model. Scatter diagram and correlation coefficients are still widely used by researchers to report the performance of their models.

Research work done during 1980's and 1990's led to the development of several performance measures (model bias, fractional bias, normalized mean square error, correlation coefficient, geometric mean bias, geometric mean variance, factor-of-two) to evaluate the air quality models. The U.S. EPA has laid some guidelines in order to validate and calibrate models in a comprehensive manner (U.S. EPA, 1992b). Note that air quality scientists and engineers do not use all the performance measures.

The toxic release models are constantly being evaluated using above statistics (Gudivaka and Kumar (1989), Hanna et al., (1991, 1993) and Kumar et al. (1993)). The results from such studies give an idea on the usefulness of output obtained from the models. The studies indicate that no one model produce consistently good performance.


After the preparation and submission of the RMP, a considerable time will be needed to implement the plan. The work will involve ongoing training, audits, and record keeping. Since the RMP document submitted to EPA becomes a public document, a public information program will be needed to diffuse the public concerns about safety. Good pubic relation will develop public trust and enhance community relations.


The deadline for the RMP submission is June 21, 1999. This paper provides an overview of the RMP program. The development of the program includes hazard assessment, prevention program, environmental emergency-response program and overall management system. It is hoped that your plant will be able to develop the plan in time.


    AIChE, "Guidelines for Evaluating the Characteristics of Vapor Cloud Explosions, Flash Fires, and BLEVEs," American Institute of Chemical Engineers (AIChE), 1994.
    Anonymous, "Handbook of Chemical Hazard Analysis Procedures," US Government Printing Office, Washington, D.C., 1989.
    Gudivaka, V., and A. Kumar, "An Evaluation of Four Box Models for Instantaneous Dense-Gas Releases," Journal of Hazardous Materials, Vol. 25, pp. 237-255, 1990.
    Hanna, S. R., and D. G. Strimaitis, "Workbook of Test Cases for Vapor Cloud Source Dispersion Models," Center for Chemical Process Safety, AIChE, 1989.
    Hanna, S. R., D. G. Strimaitis and J. C. Chang, "Evaluation of Fourteen Hazardous Gas Models with Ammonia and Hydrogen Fluoride Field Data," Journal of Hazardous Materials, Vol. 26, pp. 127-158, 1991.
    Hanna, S. R., J. C. Chang , and D. G. Strimaitis, "Hazardous Model Evaluation with Field Observations," Atmos. Environ, Vol. 27A, pp.2265-2285, 1993.
    Hanna, S.R., and P.J., Drivas, "Guidelines for use of Vapor Cloud Dispersion Models," Center for Chemical Process Safety of the American Institute of Chemical, Second Edition, 271 pp., 1996, (Note: The first edition was published in 1987).
    Kumar, A., J. Luo and G. Bennett, "Statistical Evaluation of Lower Flammability Distance (LFD) using Four Hazardous Release Models," Process Safety Progress, 12(1), pp. 1-11, 1993.
    Kumar, A., and H. G. Rao, "Software for Regulatory Compliance of Chemical Hazards," Environmental Progress, Vol. 9, No. 4, pp. m7 - m9, 1990.
    Kumar, A., and S. Vashisth, "Software for Emission Rate Modeling of Accidental Toxic Releases", American Academy of Environmental Engineers, 1999 (in press).
    U. S. Environmental Protection Agency, "Guidelines on Air Quality Models (Revised)," EPA - 450/4-78-027R, (NTIS PB 86 -245248), 1986.
    U.S. Environmental Protection Agency, "A Workbook of Screening Techniques for Assessing Impacts of Toxic Air Pollutants," EPA-450/4.88-089, Sept. 1988.
    U. S. Environmental Protection Agency, "Workbook of Screening Techniques for Assessing Impacts of Toxic Air Pollutants (Revised)," EPA - 454/R-92-024, 1992a.
    U. S. Environmental Protection Agency, "Protocol for determining the best performing model," EPA-454/R-92-025, (NTIS PB93-226082), 1992b.
    U. S. Environmental Protection Agency, "Guidance on the Application of Refined Dispersion Models to Hazardous/Toxic Air Pollutant Releases," EPA - 454/R-93-002, 1993.
    U. S. Environmental Protection Agency, "User's Guide to TSCREEN: A Model for Screening Toxic Air Pollutant Concentrations," EPA - 450/4-90-013, 1994.
    U. S. Environmental Protection Agency, "RMP Offsite Consequence Analysis Guidance," Amendment to 40 Code of Federal Regulations Part 68, Washington D.C., May 1996.