BOB VENUGOPAL is a process safety specialist with Chilworth Technology,

Inc. (250 Plainsboro Rd., Bldg #7; Plainsboro, NJ 08536; Tel: (609) 799-4449
Ask the Experts Fax: (609) 799-5559; E-mail: [email protected]). He is responsible
for running the chemical reaction hazards laboratory and assessing the ther-
mal safety of chemical processes and safe storage of reactive materials.

Thermal Venugopal is also a consultant, an expert witness , and an instructor for safe-
ty-related topics and investigations. Prior to employment at Chilworth, he was
process engineer at Novartis and FMC. Venugopal has BS, MS and PhD

Safety degrees in chemical engineering from Indian Institute of Science in Bangalore

Q. How can the rates of heat generation and heat loss with cous fluids). In the case of storage of solid material in a drum,
the surroundings be used to determine the thermal stability the temperature is highest in the center, and the temperature at
of a material under storage conditions? the surface is the same as the outside temperature.
Self-heating due to slow thermal degradation during To estimate the characteristic time of cooling, also known as
long-term storage or transportation of bulk materials repre- the critical heat dissipation time or TMRcrit (h), for situations
sents a hazard that is often underestimated. Over longer where convective heat transfer is predominant:
periods of time, even very small reaction rates may thermal-
ly self-accelerate to a runaway situation, especially where TMRcrit = 4MCp/(hA × 3,600) (2)
large quantities of material are involved. Secondary events
of high severity may be induced by a resulting thermal where M is the mass of the contents (kg), h is the heat-transfer
explosion, which can lead to a detonation and/or spreading coefficient (W/m2K) and A is the heat-transfer area (m2).
of toxic or flammable chemicals. For situations where conductive heat transfer is predomi-
A thermal explosion occurs if the rate of heat production is nant, TMRcrit (h) is calculated as:
greater than the rate of heat dissipation, or, in other words, if
the time to the maximum rate adiabatic of heat generation is TMRcrit = 0.3r2ρCp /(λ × 3,600) (3)
shorter than the characteristic time for heat loss, which can vary
from 1/2 h for a 1-m3 stirred vessel to 5 mo for a solid in a where r is the radius of the container (m), ρ is density (kg/m3),
25-m3 container. In this article, the concepts of time to maxi- and λ is thermal conductivity (W/m-K).
mum rate of heat generation and the characteristic time of cool- A runaway reaction cannot occur if TMRad is greater than
ing for reactions will be discussed for the cases where the tem- TMRcrit. TMRad is strongly dependent on temperature; where-
perature is homogeneous (e.g., stirred vessels) and not as TMRcrit is less dependent on temperature. A change in the
homogeneous (e.g., non-stirred vessels like drums and totes). outside temperature, reduction of M at constant A, or increase
Experimental methods used to characterize decomposition in A at constant M can minimize the thermal consequences.
reactions (DRs) will also be presented. The following techniques are used to characterize DRs:
When heat accumulates, DRs run very slowly at the han- Differential scanning calorimetry (DSC). This
dling temperatures. This is mainly due to poor heat transfer, microthermal analytical tool uses only milligrams of materi-
which often occurs in large, unstirred masses. Cases have al and can be run in a dynamic or isothermal mode. From the
been reported in which heat generated at low rates at the isothermal runs, q0, and Ea can be found and used in Eq. 1 to
desired storage temperature by a slow DR could not be trans- calculate TMRad.
ferred to the surroundings because the vessel was operating Adiabatic storage test. This test is conducted in an adiabat-
under nearly adiabatic conditions. As a result, the DR was ic Dewar apparatus comprising an automated 1-L reactor
accelerated beyond control, due to self-heating. encased within an adiabatic shield. Simulating reactors up to
The adiabatic temperature rise, ∆Tad, and the time to maxi- 25 m3, it determines the rate of heat generation, q0, as a func-
mum rate, TMRad, can be used to characterize the severity and tion of temperature, T0 — data that are used to calculate
probability for the worst-case scenario of a thermal runaway. TMRad. TMRad is then compared to the heat loss data relating
For zero-order reaction kinetics, TMRad (h), is calculated as: to the package or container in which the material is stored to
determine the maximum safe storage temperature.
TMRad = CpRT02/(q0Ea × 3,600) (1) Basket test. This test, which is run in an isoperiobolic
mode (i.e., where the temperature of the surroundings is
where Cp is the specific heat (J/kg-K), R is the universal gas constant), generates data that are used to determine T0, q0
constant (8.314, J/mol-K), T0 is the absolute temperature or and Ea, which are used to calculate TMRad. TMRad may be
minimum onset temperature exothermic activity of a dust compared with TMRcrit (per Eq. 2 or 3) to determine the
sample (K), q0 is the heat output at T0 (W/kg) and Ea is the material’s maximum safe storage temperature. Further, at
activation energy (J/mol). The concept of TMRad assumes least three basket sizes are used, so one can extrapolate to
that no heat is being transferred. But in real life situations, find the maximum safe operating/storage temperature for
there is always heat loss to the surroundings. any vessel size or geometry.
Heat removal also can be characterized by time. There are
two situations to consider — (a) exothermically reacting sys- Further Reading
tems where convective heat transfer is predominant (e.g., reac-
tors with mixers, reactors containing a low-viscosity liquid) 1. United Nations, “Recommendations on the Transport of Dangerous Good –
Manual of Tests and Criteria,” 3rd Ed. (1999).
and (b) exothermically reacting systems where conductive heat
transfer is predominant (e.g., storage of solid material in a 2. Expert Commission, Swiss Chemical Industry, “Thermal Process
Safety,” Booklet 8, Lucerne, Switzerland (1993).
drum, reactors that are not mixed and reactors containing vis-

16 www.cepmagazine.org December 2004 CEP