Thermal Stability Parameters
Many engineers in the process safety industry are familiar with SADT and TMR. SADT, or Self-Accelerating Decomposition Temperature, is defined as the lowest environmental temperature at which the center of the material within its packaging heats to a temperature 6°C greater than the environmental temperature after a lapse of a seven day period or less [1]. TMR, or Time to Maximum Rate, is the amount of time, given an initial temperature, that it takes for a material to reach its maximum temperature rise rate [2]. SADT is typically used in transportation for determining when a self-reactive substance needs to be shipped under sub-ambient conditions, and it takes into account heat transfer to the surroundings. TMR is fairly common when the required storage temperature of a self-reactive substance is being determined given a specific time span, and it assumes an adiabatic state.
Both SADT and TMR are popular calculations that Fauske & Associates offers, and both have very specific conditions included in their definitions as stated above. An alternative calculation, which takes process cooling into account, is the Temperature of No Return (TNR). TNR is defined as “the temperature at which the rate of heat generation of a reaction is equal to the maximum rate of cooling available” [3]. TNR goes hand in hand with thermal runaway, in that a thermal runaway is what immediately follows the TNR if no mitigating measures are in place, as shown in Figure 1.
Why TNR Matters
How does a company determine their process temperature and safety limits? It all comes down to risk tolerance and profitability – in other words, outputting the most product in the least time possible while still keeping things in a safe state with acceptable risk. While everyone would love the simple answer of an “onset temperature,” it is unfortunately more complex than a single number. Unlike flammability properties that have distinct boundaries, like flash points and autoignition temperatures, there is no temperature that acts as a “switch” to start a Figure 1: Heat Generation vs Temperature chemical reaction. They are always occurring, even if they are occurring at an imperceptible rate that cannot be detected by the calorimeter used to obtain the onset temperature. Thus, onset temperature changes for every instrument and sensitivity [4]. Kinetics dictate the reaction rates and need to be part of the assessment. Onset temperature alone should not be the defining factor for a safe process temperature.
While SADT is a useful metric, it is better suited for longer time scales. As previously stated in the introduction, SADT looks at a seven-day time span, which may not be the most applicable parameter for a manufacturing process that is completed in a matter of hours. TMR is a simpler calculation, however it assumes an adiabatic state with no benefit from the cooling system and is considered a conservative parameter. With the goal being a profitable, yet safe process, TNR is a great characteristic with which to work. It allows for a direct comparison between the heat generated by a specific reaction and the heat removal of the surrounding atmosphere/cooling system, which negates the necessity of an onset temperature. The goal of a TNR is not to determine exactly what temperature the reaction begins, but rather to explore the reaction kinetics to determine the point at which the heat removal will be overcome by the heat generation.
Incidents in the Past
Thermal runaway incidents in the past have often been the result of an overwhelming amount of heat being generated in a vessel. This is oftentimes due to an overcharge of a reactant, accumulation of a reagent, loss of cooling, loss of mixing, etc. However, one instance of thermal runaway stands out that was investigated by the U.S. Chemical Safety and Hazard Investigation Board (CSB). In April of 1998, an incident occurred at Morton International, Inc in Paterson, NJ that involved a very simple issue: Morton didn’t know that the decomposition temperature of their product was close to their upper operating temperature [5].
The production process in question was that of Automate Yellow 96 Dye. This dye was produced by mixing two chemicals, ortho-nitrochlorobenzene and 2-ethylhexylamine, and heating the mixture to initiate the reaction. Once the target temperature was reached, cooling was enabled and manually controlled by operators. This process had been run for nearly a decade at this specific site, and the operators had years of experience. They had encountered occasional temperature excursions during the desired reaction; however, management did not raise the alarm since the temperature always returned to within the operating limits.
There were several key findings detailed in the report. For example, regarding the vessel itself, Morton switched from a 1000-gallon kettle to a 2000-gallon kettle and increased the overall batch by approximately 9%. This was successfully run for a few years, albeit with an increased number of temperature excursions (that were, once again, not investigated or addressed). The emergency relief devices were sized for a nonreactive fire scenario with a different component altogether, and there were no emergency cooling systems in place. Overall, Morton did not have the safety systems in place, nor did they have sufficient knowledge of the potential thermal hazards to run the Yellow 96 Dye synthesization process.
The process was fairly simple. Once all of the materials were in the kettle, procedures stated that the batch should be heated to 90°C to initiate the reaction, and then the temperature was to be maintained between 90°C and 110°C. “To hold a constant temperature in the kettle after the synthesis reaction had begun, the operators manually adjusted the flow of cooling water through the water jacket, trying to balance the heat removal rate with the heat generation rate. Without any indicators for water-flow rate or for cooling-water inlet and outlet temperatures, the operators controlled the kettle’s temperature based on in-house training and experience” [5]. There were multiple occasions where the temperature surpassed the 110°C operating limit, however there was no concern for a runaway because the potential for a runaway reaction was unknown at the time. There was only a concern for the quality of the product, as heating above 160°C would decrease the yield and quality [5].
What the engineers were unaware of was that there was a second unintended reaction that consisted of the decomposition of Yellow 96 with an observed onset of 195°C, as was determined during the CSB’s investigation [5]. All it took was one batch heating a little too much and suddenly the heat generation overwhelmed the kettle’s cooling capacity. The cooling water valves to the jacket were confirmed open by multiple operators, and the same operators confirmed in interviews later that they could hear the sound of water flowing through the piping. However, the self-heating rate continued to increase due to the cooling capacity being surpassed, and the batch temperature shot past the decomposition onset temperature. The vessel overpressurized shortly after and exploded.
The decomposition reaction was what led to the loss of containment, but the thermal runaway initially began when the first intended reaction started generating heat too quickly for the cooling to mitigate. If a maximum operating temperature below the TNR had been in place, this incident may have been avoided.
Determining TNR With Adiabatic Calorimetry Data
There are multiple ways that the TNR can be determined, including performing calculations with known material properties to determine the heat generation of the reaction and the cooling capacity of the cooling system. But what happens when there isn’t enough information known about the reaction? Adiabatic calorimetry testing is a great option. A low phi-factor test that provides adiabatic temperature rise rate data is vital in getting an accurate heat generation estimate. It’s important to note that heat losses decrease significantly as the size of the process is scaled up. The phi-factor must be accounted for to know the rate of heat generation at the process scale and accurately determine the TNR.
However, getting the adiabatic calorimetry data is not sufficient on its own. Each specific vessel, reactor, storage container, etc. must be evaluated for its cooling capacity. Some common questions are: Is ambient temperature cooling being relied on? What is the ambient temperature? If there is a cooling system, at what capacity is it being used? Can temperature excursions above the normal operating temperature be handled? Perhaps most importantly, has the cooling system been evaluated for the worst-case scenario?
Similar to the plot shown in Figure 1, the TNR can be estimated for varying concentrations/masses by finding the intersection between the temperature rise rate line and the heat removal rate for different vessels or storage containers. This is shown in Figure 3, with the example using ARC data that has been phi-corrected [6]. Multiple containers are compared, and there are significant differences in the intersecting points between the TNR values for each one.
Conclusion
There are several determinable properties that can assist in maintaining safe manufacturing conditions. While SADT and TMR are great pieces of information, they are more applicable to thermal stability concerns or can lead to overly conservative operating limits given TMR does not take heat loss to its surroundings into account. When it comes to determining the maximum temperature a process can be allowed to reach before a thermal runaway could become uncontrolled, TNR is an important value to determine.
Incidents can be caused by many possible upsets or failures. However, knowing the temperature at which a reaction’s heat generation will overwhelm the system’s cooling capacity creates a threshold temperature that should never be reached. If the TNR is surpassed, every operator should know that emergency measures need to be enacted and that process relief devices may require activation. In addition, emergency relief devices should be properly sized to safely relieve the pressure from a runaway reaction and prevent vessel failure or loss of containment. Part of the Morton incident was a delay in response to a thermal runaway, mostly because they were unaware of the chance of a secondary reaction. It is vital that safety parameters are in place for any and all processes, and that the properties of the components, their products, and any intermediates or side products are investigated, well-understood, and documented.
References:
- Kurko K., Kozlowski C., Roduit B. Consideration of Autocatalytic Behavior in Determination of Self Accelerating Decomposition Temperature. Presented at: 39th Annual Conference of NATAS, 2011.
- Ott B., Welchert N., Delafontaine L., Frajnovi M., Reza A. Considerations for the Safe Handling and Processing of Unstable Materials. Chemical Engineering Progress, An AiChE Publication; February 2025. Available at https://www.aiche.org/resources/publications/cep/2025/february/considerations-safe-handling-and-processing-unstable-materials.
- Chemical Reaction Hazards, 2nd Edition Edited by J. Barton and R. Rogers. Institute of Chemical Engineers: Rugby, U.K. 1996. 225 pp. ISBN 0-85295-341-0.
- Fauske & Associates. What is an Onset Temperature, and How Should I Use it to Better Understand My Reactive Hazards? Fauske. December 15, 2023. Available at: https://www.fauske.com/blog/what-is-an-onsettemperature-and-how-should-i-use-it-to-better-understand-my-reactive-hazards
- U.S. Chemical Safety and Hazard Investigation Board. Morton International Inc. Runaway Chemical Reaction Investigation Report (Report No. 1998-06-I-NJ). Available at: https://www.csb.gov/morton-international-inc-runaway-chemical-reaction/.
- Horsch S. A Reactive Chemicals Workshop: Practical Considerations of Thermodynamic and Kinetic Information from Calorimetry for Safe Chemical Operations at All Scales. Presented at: 2025 Purdue Process Safety & Assurance Center, May, 2025.