Highly reactive industrial chemicals have made available an enormous variety of useful products that we often take for granted. However, chemical reactivity can present significant hazards under certain conditions, and the control of chemical energy continues to be a significant concern for facilities that process, handle, transport or store chemicals. The potential to inflict harm or damage exists because reaction energy can cause over-pressurization of process equipment and subsequent release of toxic or flammable materials. To control reaction energy and manage reactive hazards and risk, it is necessary to characterize chemical reactivity that can appear under normal and likely upset conditions. Certainly, general hazard-assessment methods can be used, but they usually require extensive resources. This article describes how to adequately evaluate reactivity with limited resources.
The U.S. Chemical Safety and Hazard Investigation Board (CSB) in 2002 released a report on 167 reactive-chemical incidents that occurred in the U.S. between January 1980 and June 2001 . More than 50% of these incidents involved chemicals that are not covered by U.S. Occupational Safety and Health Administration (OSHA) or U.S. Environmental Protection Agency (EPA) regulations. Existing OSHA and EPA regulations are based on lists of individual chemicals that are considered inherently unstable or toxic. However, reactive hazards are associated with an enormous range of chemical reactivity behavior based on wide ranges of process and upset conditions that challenge attempts at regulation.
A complex issue
The intrinsic properties of a chemical are based on the molecular electronic structure as represented by its molecular formula. Although each chemical has unique properties that affect reactive behavior, observed reactivity and, therefore, reactivity hazard is not an inherent property. In fact, the observed reactivity results from the interaction of the chemical in contact with other chemicals and materials at the conditions of the interaction.
Because there are endless combinations of chemicals and conditions, no simple test, indicator or assessment can fully characterize a reactivity hazard. Therefore, lists of reactive chemicals are futile and misleading because they can suggest that an unlisted chemical is never hazardous. A listed reactive chemical can be benign under some conditions, while a normally benign compound might not be, as was the case with water, which in the presence of methyl isocyanate, led to the Bhopal tragedy. So, it is not surprising that a thorough reactive hazard assessment can be expensive, time consuming and require extensive expertise. Because of limited resources, a sufficient reactivity assessment often is not performed to characterize hazards or to determine whether the process must be altered to decrease the risk to acceptable levels. However, the initial steps of a reactivity assessment to identify basic hazards are critically important to reduce risk, and these steps require minimal resources.
Reactivity hazard assessment is best accomplished using a multi-leveled approach  that begins with available information about the chemicals, including records of previous incidents, and proceeds with computational and experimental tests (Table 1), which indicates relative experience, time and cost, and gives examples for the various levels.
<Production dept., place table close to here>
Level 1 prescreening
Level 1 prescreening assembles information needed for a hazard assessment and anticipates potential hazards. Material Safety Data Sheets (MSDSs) are always available but often are incomplete and should be supplemented with other sources to obtain reliable information, such as physical and thermodynamic properties, reactivity, flammability and toxicity, on each chemical. Also important is compatibility information, which is available from a variety of sources; for instance, the U.S. National Oceanic and Atmospheric Administration offers a Chemical Reactivity Worksheet (CRW) program for each chemical pair of more than 6,000 chemicals. (http://response.restoration.noaa.gov/chemaids/react.html). Many reference books and literature articles provide free of charge MSDSs as well as other Level 1 information. Brethericks Handbook of Reactive Chemical Hazards, for example, is available in libraries and includes numerous records of incidents that are valuable for anticipating potential reactivity hazards.
Chemical bonds and structures. As part of prescreening, a look at the molecular structure, the fundamental basis for intrinsic properties, is useful for anticipating potential hazards. Some chemical bonds and structures, such as unsaturated bonds (acetylene, butadiene) that are not in benzene rings, are more likely than others to exhibit high reactivity or instability. Other examples include certain compounds containing nitrogen-nitrogen bonds (azo compounds, azides) or oxygen-oxygen bonds (peroxides, peroxyacids) and certain structures containing metals (metal fulminates, arylmetals) or halogens (halamides, halogen azides). Compounds with strained rings, such as ethylene oxide, are less stable and have a greater tendency to react to a more stable state with a release of energy. Extensive examples of bonds and structures are provided in Brethericks , the Center for Chemical Process Safety (CCPS) Guidelines  and Lees Loss Prevention .
Chemical reaction types. It also is important to gather information on the types of expected reactions and their behavior, which include oxidation, oxidation-reduction, polymerization, combustion, decomposition, rearrangement and autocatalytic [4, 6].
Physical processes. Heat generation or consumption by such processes can significantly affect system temperature and, therefore, chemical reaction rate. Examples of such physical processes that are part of Level 1 information are melting, vaporization, wetting, adsorption, hydration, mixing, agitation, combining, grinding and distillation.
Level 2 screening
After Level 1 information is gathered, experimental and computational screening tests should be performed to identify and characterize basic hazards, including self-reactivity or thermal stability, oxidizers, peroxide formation sensitivity, spontaneous combustibility, reactivity with air, shock sensitivity and incompatibility . A useful guide for identifying and managing basic hazards  is available free from the OSHA website on reactive chemical hazards (http://www.osha.gov/SLTC/reactivechemicals/). Even though the necessary tests incur more costs than Level 1 information searches, they still only require limited resources as part of hazard assessment to design and implement reliable chemical processes.
Included in a hazard assessment are scenarios or chains of events, developed to determine how the identified hazards can threaten process equipment and personnel. For this analysis, sensitivities of the chemicals to likely contaminants and conditions, as measured in screening tests, together with lessons learned from investigations of previous incidents involving the chemicals are extremely valuable. For each developed scenario, the potential consequence severity and the likelihood of each event are analyzed as needed to manage risk.
Experimental screening. Many calorimetric devices are available for testing reaction behavior at process and likely upset conditions. A Differential Scanning Calorimeter (DSC) is useful for examining a few milligrams of sample over wide ranges of temperature to identify exothermic processes and estimate evolved heat . Other small calorimeters, such as the ARSST , test a few grams of sample to estimate reaction behavior, including heat of reaction, and pressure change.
Computational screening. Although accurate measurements of reaction energies or heat are expensive, reaction energy can be calculated by subtracting the heats of formation of the products from those of the reactants. Experimental heats of formation are available for many molecules from the National Institute of Science and Technology (NIST) Webbook (http://webbook.nist.gov). Reaction energy can be used to develop parameters for reactivity prediction methods. An example reactivity parameter is the Computed Adiabatic Reaction Temperature (CART), which is the maximum temperature under adiabatic conditions from the heat of reaction and heat capacity of reaction components .
Especially for molecules with unusual functional groups and reactive intermediates, a calculation method is a convenient way to determine heats of formation when experimental values are not available. High-level quantum chemical calculations can rival experimental values in accuracy but can be time consuming and expensive <em dash> and often are not required. As screening tools, however, empirical and semi-empirical desktop software methods with usually acceptable accuracy are available at low cost.
The CHETAH computer program, which employs the Benson group additive method, is useful for estimating gas-phase heats of formation . MOPAC within the ChemOffice Ultra 2004 programs (http://www.cambridgesoft.com/) also calculates gas-phase heats of formation. MOPAC includes semi-empirical quantum methods, in which many of the electron interaction terms are neglected or replaced by empirical parameters based on measured data. These widely applicable methods are inexpensive and require only about a minute on a personal computer to calculate a heat of formation .
Employing a multi-leveled approach to hazard assessment, information gathering and hazard identification is the critical preliminary to hazard characterization. Each step along this assessment route should lower the overall risk. Examples of basic reactive chemical hazards from the CSB 167 incidents , summarized below, demonstrate that relatively low-cost Level 1 and 2 methods could have identified, and semi-quantitatively characterized, basic hazards that were involved in each incident.
CSB Incident No. 4, spontaneous combustibility. A fire appeared in a scrubber tank that contained mineral oil and alkyl aluminums, which are pyrophoric and ignite on contact with air or water. Although it could have been determined from a Level 1 search, the propensity of water-contaminated alkyl aluminums to self ignite and be an ignition source for mineral oil was not identified, so personnel had not prepared procedures for safe handling and use of these chemicals.
CSB Incident No. 167, oxidizer. Personnel were unclogging a dust collector that removed combustible dust, including ammonium perchlorate, a powerful oxidizing agent. The personnel were not aware of the dust-explosion initiation potential of ammonium perchlorate, which could have been discovered from a Level 1 search.
CSB Incident No. 11, thermal runaway polymerization, shock sensitivity. An overheated tank containing 80,000 gallons of acrylic acid required extensive spraying with water to prevent explosion. Acrylic acid is chemically stable under normal storage and handling conditions, but is sensitive to peroxide formation and exothermic polymerization in the presence of heat or contaminants that can deactivate peroxide formation inhibitors. Organic peroxides sensitivity to heat, shock and friction has resulted in many unexpected explosions. Also, with its relatively high freezing point of 14°C, acrylic acid often partly solidifies, with the peroxide formation inhibitor remaining in the liquid phase. Uninhibited acrylic acid polymerizes exothermically at ambient temperature and may accelerate to an explosive state. This basic information about acrylic acid, organic peroxides and potential runaway behavior could have been discovered from a Level 1 search that included records of previous incidents. Level 2 screening or higher-level tests could characterize the potential hazards and further reduce risk.
CSB Incident No. 16, reactivity with air, self reactivity, incompatibility. NaK used as thermometric fluid spilled into a furnace. Personnel were not aware that NaK reacts with air to form KO<subscript>2, which can decompose violently in the presence of hydrocarbons. During cleaning operations, mineral oil was used to wash the furnace walls, resulting in an explosion due to incompatibility with the KO<subscript>2. This reactivity and compatibility information is available from a Level 1 search.
CSB Incident No. 32, water reactivity. A process to produce alkyl benzene for detergents had been altered by replacing aluminum chloride with Al powder without going through management-of-change procedures . Level 1 information would have shown that Al powder can react exothermically with water. Following this substitution, the reactor was overfilled and the bottom outlet became fouled with solid catalyst residue. Because tests showed that the residue reacted with water, it was decided to inject ambient temperature water to dissolve the solids. Instead, personnel injected continuous steam and within minutes the reactor vessel exploded. Because the reactivity tests had been performed only at ambient temperature, the exothermic reaction of aluminum powder with hot water or steam to form hydrogen was not identified. This incident teaches the importance of management of change, a need for testing at normal and likely upset conditions, and the importance of pressure generation as part of an overall hazards assessment.
Additional explosion hazards due to combustible materials can include:
Releases of flammable gases that disperse and encounter ignition sources ;
Aerosols produced from leakage of industrial fluids to ambient pressure from higher pressures ; and
Dust accumulations that can form airborne dust clouds that disperse and explode .
Level 3 detailed reactivity tests
Screening tests satisfy or at least greatly simplify minimum requirements for a hazard assessment. In some cases based on Level 1 and Level 2 information, however, additional quantitative information using more detailed computations or experimental tests may be required to reduce risk to an acceptable level. Computational tests include quantum methods such as HF, G2 and B3LYP , while experimental methods include tests under adiabatic conditions using adiabatic calorimeters such as the APTAC or ARC . These tests require increased resources, but they can explore sensitivities of the chemicals and provide more accurate values of reaction heats and activation energies for estimating hazard evaluation parameters, such as time to maximum reaction rate, which are useful for risk quantification and reduction. The following example from the CSB 167 incidents illustrates the value of Levels 2 and 3 testing.
CSB Incident No. 1, self reactivity, thermal decomposition. Nylon polymer in a process vessel had decomposed to carbon dioxide and other gases. Because the foamy polymer fluid had blocked vessel tubing, operators opening the vessel were not aware of the trapped pressure. The partially unbolted cover blew off and the released hot fluid ignited. Following the incident, adiabatic calorimetric testing showed that the thermal decomposition of the polymer was endothermic with significant pressure generation (Figure 1) . Closed-cell testing at Levels 2 and 3 over a temperature range that included the primary and secondary decomposition reaction systems could have identified and measured pressure formation behavior, including maximum pressure and pressure generation rate. Other reactivity characterization information for hazard assessment is onset temperature for significant decomposition, the difference between the main process temperature and the decomposition onset temperature and, from Level 3, the time to the maximum pressure generation rate under adiabatic conditions.
Thermal decomposition of a nylon polymer generated significant pressure that ultimately led to a fluid release and fire.
This incident demonstrates the importance of pressure as a hazard indicator and the danger of relying on a single indicator, such as the overall reaction energy, for hazard classification.
Tools you can apply
Before hazards can be controlled, they must be identified. The multi-tiered approach and methods we have outlined can play a key role in identifying reactivity risks, without requiring extensive resources. The cited CSB incidents underscore the value of the approach. Basic reactivity hazards should have been uncovered from a Level 1 information search in all but the last incident case study. If Level 1 information had been employed, followed by Level 2 screening for hazard identification and initial characterization, the number and severity of incidents should have been greatly reduced. Level 3 tests for more detailed characterization require greater resources, but just the knowledge that additional information beyond Level 2 is needed helps to determine whether a process can be performed economically at a manageable level of risk or must be modified or abandoned. Additional information is available from many sources including the periodically updated websites of OSHA (www.osha.gov/SLTC/reactivechemicals/), and the Mary Kay OConnor Process Safety Center (http://process-safety.tamu.edu).
|Sam Mannan and William Rogers co-teach a course concerning reactive chemical hazards as part of a continuing education program provided by the Mary Kay OConnor Process Safety Center, Texas A&M University, College Station, Texas. For more information concerning available courses, call (979) 458-1863 or visit http://process-safety.tamu.edu.|
1. U.S. Chemical Safety and Hazard Investigation Board, Hazard investigation: improving reactive hazard management, Rep. No. 2001-01-H (2002). http://www.csb.gov/completed_investigations/info.cfm?ID=21.
2. Mannan, M.S., W.J. Rogers and A.A. Aldeeb, A systematic approach to reactive chemicals analysis, <itals>Proceedings of HAZARDS XVI<end itals>, p. 41, Institution of Chemical Engineers, Rugby, U.K. (2001).
3. Bretherick, L., P.G. Urben, and M.J. Pitt, Brethericks handbook of reactive chemical hazards, 6th ed., Butterworths, London, U.K. (1999).
4. Guidelines for chemical reactivity evaluation and application to process design, Center for Chemical Process Safety, New York (1995).
5. Mannan, M.S., Lees loss prevention in the process industries: hazard identification, assessment and control, Butterworth-Heinemann, Burlington, Mass. (2004).
6. Leggett, D. Chemical reaction hazard identification and evaluation: taking the first steps, <itals>Proc. Safety Prog.<end itals>, 23 (1), p. 21 (2004).
7. Wei, C., W.J. Rogers and M.S. Mannan, Application of screening tools in the prevention of reactive chemical incidents, <itals> J. Loss Prev. in the Proc. Ind.<end itals>, 17, p. 261 (2004).
8. Johnson, R.W., S.W. Rudy and S.D. Unwin, Essential practices for managing chemical reactivity hazards, Center for Chemical Process Safety, New York (2003).
9. Burelbach, J.P., Advanced reactive system screening tool, <itals>Proceedings of the Mary Kay OConnor Process Safety Center Symposium<end itals>, Texas A&M University, College Station, Texas (1999).
10. Melham, G., Reactivity screening made easy, ioMosaic Corp., Salem, N.H. (2003).
11. Shanley, E.S. and G.A. Melham, A review of CHETAH 7 hazard evaluation criteria, <itals>J. Loss Prev. in the Proc. Ind.<end itals>, 8 (5), p. 261 (1995).
12. U.S. Chemical Safety and Hazard Investigation Board, Management of change, CSB32, Rep. No. 2001-04-SB (2001). (http://www.csb.gov/safety_publications/docs/moc082801.pdf)
13. Fire protection guide to hazardous materials, 10th ed., National Fire Protection Association, Quincy, Mass. (1991).
14. Krishna, K., W.J. Rogers and M.S. Mannan, Prediction of aerosol formation for safe utilization of industrial fluids, <itals>Chem. Eng. Prog.<end itals>, 100 (7), p. 25 (2004).
15. Eckhoff, R.K., Dust explosions in the process industries, 3rd ed., Elsevier (Gulf Professional Publishing), New York (2003).
16. Cramer, C.J., Essentials of computational chemistry, Wiley, New York (2002).
17. U.S. Chemical Safety and Hazard Investigation Board, Thermal decomposition incident, CSB1, Rep. No. 2001-03-1-GA (2002). (http://www.csb.gov/index.cfm?folder=completed_investigations&page=info&INV_ID=2)
W. J. Rogers is a research scientist at the Mary Kay OConnor Process Safety Center at Texas A&M University, College Station, Texas. E-mail him at firstname.lastname@example.org.
C. Wei was a research assistant at the Mary Kay OConnor Process Safety Center, and is currently employed by DNV Consulting. E-mail her at email@example.com.
M. S. Mannan is director of the Mary Kay OConnor Process Safety Center. E-mail him at firstname.lastname@example.org.