GCPS 2015 __________________________________________________________________________

Relief System Sizing for Runaway Chemical Reactions: A Simple

Comprehensive Approach

Ken Kurko
Fauske & Associates, LLC
16W070 83rd Street
Burr Ridge, IL 60527
kurko@fauske.com

Prepared for Presentation at

American Institute of Chemical Engineers
2015 Spring Meeting
11th Global Congress on Process Safety
Austin, Texas
April 27-29, 2015

AIChE shall not be responsible for statements or opinions contained

in papers or printed in its publication
GCPS 2015 __________________________________________________________________________

Relief System Sizing for Runaway Chemical Reactions: A Simple

Comprehensive Approach

Ken Kurko
Fauske & Associates, LLC
16W070 83rd Street
Burr Ridge, IL 60527
kurko@fauske.com

Keywords: Vent sizing, runaway reactions, source terms, adiabatic calorimetry, flow regime

Abstract
When conducting relief sizing calculations for reactive systems, experimental data is of
paramount importance. Experimental techniques for gathering the necessary kinetic data
highlighting VSP2 (Vent Sizing Package 2) and ARSST (Advanced Reactive System Screening
Tool) testing will be outlined. In addition, the principles behind the system classifications (vapor,
gassy and hybrid) will be outlined with example experimental data sets. Relief sizing
calculations and source term considerations for each system will be discussed. Two-phase flow
and experimental techniques for determining flow regime (vapor, churn, bubbly or homogeneous
flow) will be presented with the goal of providing the audience with a simple comprehensive
experimental approach to designing relief systems.

Upset Scenario Selection

Evaluation of reactive chemistries for relief system design requires consideration of plausible
upset scenarios. What exactly goes wrong (and when it goes wrong) in a chemical process can
have a great affect on the rates at which a reaction runs away. This in turn affects relief
requirements and whether or not a particular relief system design is adequate to handle the worst
potential scenario. Common upset scenarios include, but are not limited to, loss-of-cooling
during normal process conditions, inadvertent heating, wrong reagent addition, and addition at
wrong temperature. Upset scenario selection is typically done in a process hazard analysis
(PHA) that includes process engineers, chemists, reactive hazards experts, operators, and
maintenance workers.

Experimental Techniques
Once plausible upset scenarios have been determined, it is critical that adiabatic calorimetry
experiments be performed to characterize the reactive chemistry in question. The VSP2 and
ARSST adiabatic calorimeters provide a means to directly simulate a specific upset scenario on
the bench scale and do so in a safe manner. Due to the low phi-factor, the measured reaction
rates are directly applicable to the process scale.
GCPS 2015 __________________________________________________________________________

Source Term Considerations

During a reactive upset scenario, overpressurization of a process vessel can be caused by an
increase in vapor pressure, generation of non-condensable gas, or a combination of the two.
These overpressure sources are classified as vapor, gassy, or hybrid. Adiabatic calorimetry
experiments provide a means for determining the source of overpressure.

Reactive systems in which the increase in pressure is due solely to an increase in vapor pressure
are commonly referred to as vapor systems. Vapor systems have the advantage that the latent
heat from boiling prevents the temperature from increasing (or tempers the system). As a result,
the temperature rise due to the runaway reaction can be controlled by venting. In these cases, the
relief system must be designed to accommodate the vapor generated during venting. As such,
the relief requirement is based on the temperature rise rate. Adiabatic calorimetry experiments
for these types of systems are typically run as closed cell tests. This allows for vapor pressure
data to be collected in addition to reaction rate data.

Reactive systems in which the increase in pressure is due solely to the generation of non-
condensable gas are commonly referred to as gassy systems. When dealing with gassy reactive
systems, it is important to recognize that the reaction temperature rise cannot be controlled by
venting. Because the temperature cannot be controlled in these cases, the relief system must be
designed to accommodate the peak rate of non-condensable gas generation. Because obtaining
the peak pressure rise rate is important, adiabatic calorimetry experiments for these types of
systems are commonly run as open cell tests to allow for a large volume of non-condensable gas
formation.

Reactive systems in which the increase in pressure is due to vapor and non-condensable gas
generation are commonly referred to as hybrid systems. Latent heat is available to remove
reaction heat. Due to the generation of both vapor and non-condensable gas, it is important to
consider both temperature and pressure rise rates for relief design. Similar to gassy systems,
adiabatic calorimetry experiments for these types of systems are often times run as open cell
tests. This not only provides a large volume for non-condensable gas formation but also allows
for confirmation of tempering.

Flow Regime Considerations

After the determinations of system classification and reaction rates, it is also important to
consider vessel flow regime. The flow regime determines the amount of two-phase flow that
will occur during venting. This in turn affects the size of the required emergency relief vent.
Predictions of the flow regime based on material properties of the vessel contents are not
reliable. Prior industrial experience with the chemistry in question or experimental data are
required.

Either adiabatic calorimeter previously mentioned can be used for flow regime determination.
The ARSST utilizes a Flow Regime Detector (FRED) for distinction between foamy or non-
foamy venting. The FRED uses a heater and thermocouple combination in the headspace of the
test cell. The heater is programmed to heat the headspace above the test cell to a temperature
GCPS 2015 __________________________________________________________________________

hotter than the sample temperature. Foamy conditions are detected when the thermocouple in
the headspace is quenched by the sample. Specially designed tests for flow regime
determination in the VSP2 are called blowdown tests. A blowdown test involves depressurizing
the test cell and determining the quantity of material remaining afterwards. The small scale
VSP2 test is designed to mimic the dimensionless superficial velocity that would be seen on the
large scale during venting. This way, one can make the distinction between churn-turbulent
(non-foamy) or bubbly and homogeneous (foamy) flow regimes.

Fauskes Simple Relief Sizing Equation

Fauske has worked to develop a simplified relief sizing methodology for reactive systems that
can be used with low phi-factor adiabatic calorimetry data and limited knowledge of material
properties. This methodology bases the required vent size on the all vapor or gas vent area by
allowing for a modest overpressure. While the vent area is calculated on this basis, it does not
imply that two-phase flow will not occur. The following equation can be applied to all
classifications of reactive systems venting at or above the critical pressure ratio.

1 m c p T R T
1/ 2
m v P MWg
1/ 2

A =
0.61 C D P MWv m t P R T

where A = required vent area (m2)

CD = discharge coefficient (-)
m = mass of vessel contents (kg)
cp = heat capacity of vessel contents (J/kgK)
T = temperature rise rate from calorimetry test (K/s)
= latent heat of vessel contents (J/kg)
P = venting pressure (Pa)
R = universal gas constant (8314.47 J/kmolK)
T = venting temperature (K)
MWv = molecular weight of vapor (kg/kmol)
v = freeboard volume of test (m3)
P = pressure rise rate from calorimetry test (Pa/s)
mt = mass of test sample (kg)
MWg = molecular weight of gas (kg/kmol)

In order to use the aforementioned equation for vapor systems, the overpressure available during
venting is required to be at least 40% on an absolute basis. The temperature, pressure,
temperature rise rate, and material properties for vapor systems are evaluated at the set pressure
of the relief device. It is also important to note that the equation is applicable to non-foamy
systems. If a foamy flow regime is expected to prevail, the calculated vent area should be
increased by a factor of 2.

For gassy systems, the temperature, pressure, and material properties are evaluated at the
maximum allowable accumulation pressure (MAAP). Applicable code permits an overpressure
GCPS 2015 __________________________________________________________________________

of 10% above the maximum allowable working pressure (MAWP) of the vessel on a gauge basis.
The pressure rise rate is the maximum measured during the adiabatic calorimetry test.

For tempered hybrid systems, the temperature, pressure, temperature rise rate, pressure rise rate,
and material properties are evaluated at the set pressure of the relief device.

Vapor System Example

3500 kg of a phenol-formaldehyde resin is produced in a reactor with a volume of 5 m3 and a
maximum allowable working pressure of 30 psig. The reaction is run at a temperature of 50C
with a sodium hydroxide catalyst being metered to a mixture of phenol, water, and
formaldehyde. The total batch has a final composition as follows:

Concentration
Component
(wt.%)
Phenol 31.25
Water 20.50
50% Formaldehyde 44.25
50% Sodium Hydroxide 4.00

Results from a PHA indicate an abnormally fast addition of the catalyst charge would overwhelm
the cooling system of the reactor leading to a runaway reaction. It is desired to size a rupture
disk with a set pressure of 10 psig based on this upset scenario.

Phenol-formaldehyde systems such as this are known to be foamy vapor systems, making the
VSP2 an excellent choice for obtaining runaway reaction data. A closed-cell VSP2 test was run
to simulate this upset scenario. The phenol, water, and formaldehyde were loaded to the test cell
(in the appropriate ratios) and heated to a temperature of 50C. The sodium hydroxide catalyst
was then injected to the test cell and the reactants were allowed to run away adiabatically. The
data from this test that are pertinent to the vent sizing evaluation are provided in Figures 1 and 2.
GCPS 2015 __________________________________________________________________________

Figure 1 Pressure vs. Temperature for Phenol-Formaldehyde

Figure 2 Temperature Rise Rate vs. Temperature for Phenol-Formaldehyde

GCPS 2015 __________________________________________________________________________

In this case, the pressure within the test cell closely matches the vapor pressure of water at
conditions where relief is expected to occur. Since there is an abundance of water in the batch,
this reactive system is expected to be a vapor system during venting. Based on Figure 1, the
desired relief set pressure of 10 psig corresponds to a temperature of 115C. The corresponding
temperature rise rate of 23.1C/min is shown in Figure 2. The vent sizing calculation for this
particular upset scenario is shown below. A factor of 2 has been used to account for the
expected foamy behavior of the phenol-formaldehyde system. In addition, a discharge
coefficient of 0.5 has been chosen to represent reasonably designed outlet piping from the
rupture disk.

2 m c p T R T
1/ 2

A =
0.61 C D P MWv

1/2
2 (3500 kg)(3380 J/kg K)(0.385 K/s) (8314.47 J/kmol K)(388 K)
A =
0.61 0.5 (2.2 10 6 J/kg)(1703 00 Pa) (18.015 kg/kmol)

A = 3.3710-2 m2 or 52.3 in2

Gassy System Example

210 kg of 40% dicumyl peroxide in 2,2,4-trimethyl-1,3-pentanediol diisobutyrate is stored in a
340 L tank. The maximum allowable working pressure of the tank is 80 psig.

Results from a PHA indicate a fire in the surrounding area would elevate the temperature of the
tank to cause decomposition of the dicumyl peroxide. The external fire exposure rate was
determined to be 0.5C/min. It is desired to size a rupture disk with a set pressure of 50 psig
based on this upset scenario.

Because non-condensable gas is expected to be generated due to decomposition of the dicumyl

peroxide, the ARSST is an excellent choice for obtaining runaway reaction data. An open-cell
ARSST test was run to simulate this upset scenario. The dicumyl peroxide and diisobutyrate
were loaded to the test cell (in the appropriate ratios) and heated at an external rate of 0.5C/min.
In accordance with open cell methodology, the test was performed with an initial nitrogen
backpressure equal to the storage tank MAAP of 88 psig. This ensures sufficient reaction rate
data are obtained for vent sizing purposes. The data from this test that are pertinent to the vent
sizing evaluation are provided in Figures 3 and 4.
GCPS 2015 __________________________________________________________________________

Figure 3 Pressure vs. Temperature for 40% Dicumyl Peroxide in Diisobutyrate

Figure 4 Pressure Rise Rate vs. Temperature for 40% Dicumyl Peroxide in Diisobutyrate
GCPS 2015 __________________________________________________________________________

In this case, a significant amount of non-condensable gas was generated as evidenced by the cool
down pressure of 103 psig. The peak reaction temperature from the runaway reaction was
250C. Since the diisobutyrate does not boil until a temperature of 280C, significant vapor
generation will not occur. This reactive system is expected to be a gassy system during venting.
Based on Figure 4, the maximum pressure rise rate is 477 psi/min and occurs at a temperature of
230C. The vent sizing calculation for this particular upset scenario is shown below and based
on all gas venting. A discharge coefficient of 0.5 has been chosen to represent reasonably
designed outlet piping from the rupture disk.

1 m v P M Wg
1/ 2

A =
0.61 C D m t P R T

1/2
1 (210 kg)(3.5 10 -4 m 3 )(54800 Pa/s) (44.01 kg/kmol)
A =
0.61 0.5 (8 10 kg)(708090 Pa)
-3
(8314.47 J/kmol K)(503 K)

A = 7.5610-3 m2 or 11.7 in2

Hybrid System Example

2000 kg of 25% hydrogen peroxide is stored in a 700 gallon tank. The maximum allowable
working pressure of the tank is 100 psig.

Results from a PHA indicate iron contamination could cause a runaway reaction due to
accelerated decomposition rates of the hydrogen peroxide. It is desired to size a rupture disk
with a set pressure of 20 psig based on this upset scenario.

Because non-condensable gas is expected to be generated due to decomposition of the hydrogen

peroxide, open cell VSP2 testing is an excellent choice for obtaining runaway reaction data. The
hydrogen peroxide and iron contaminant were loaded to the test cell (in the appropriate ratios)
and subjected to heat-wait-search methodology. In accordance with open cell methodology, the
test was performed with an initial nitrogen backpressure equal to the storage tank MAAP of 110
psig. This ensures sufficient reaction rate data are obtained for vent sizing purposes. The data
from this test that are pertinent to the vent sizing evaluation are provided in Figures 5 through 7.
GCPS 2015 __________________________________________________________________________

Figure 5 Pressure vs. Temperature for 25% Hydrogen Peroxide System

Figure 6 Temperature Rise Rate vs. Temperature for 25% Hydrogen Peroxide System
GCPS 2015 __________________________________________________________________________

Figure 7 Pressure Rise Rate vs. Temperature for 25% Hydrogen Peroxide System

In this case, a significant amount of non-condensable gas was generated as evidenced by the cool
down pressure of 133 psig. The peak temperature during this test was 179C. This implies that
the water will boil during a venting scenario and tempering of the reaction is possible. In order
to verify this assumption and determine at what temperature tempering may occur, an open-cell
test was run at 20 psig. The data from this test that are pertinent to the vent sizing evaluation are
provided in Figure 8.
GCPS 2015 __________________________________________________________________________

Figure 8 Comparison Temperature Rise Rates for 25% Hydrogen Peroxide System

As evidenced in Figure 8, the low backpressure test exhibited tempering at a temperature of

124C. This reactive system is thus expected to be a hybrid system during venting. Based on
Figures 6 and 7, the reaction rates corresponding to the tempering point are 21C/min and 6
psi/min. The vent sizing calculation is shown below and based on all gas and vapor venting. A
discharge coefficient of 0.5 has been chosen to represent reasonably designed outlet piping from
the rupture disk.

1 m c p T R T
1/ 2
m v P MWg
1/ 2

A =
0.61 C D P MWv m t P R T

1/2
1 (2000 kg)(3900 J/kg K)(0.35 K/s) (8314.47 J/kmol K)(397 K)
A =
0.61 0.5 (2.2 10 6 J/kg)(2392 48 Pa) (18.015 kg/kmol)

1/2
1 (2000 kg)(3.8 10 -3 m 3 )(689 Pa/s) (32.0 kg/kmol)

0.61 0.5 (8.0 10 kg)(239248 Pa)
-2
(8314.47 J/kmol K)(397 K)

A = 1.0110-2 m2 or 15.6 in2

GCPS 2015 __________________________________________________________________________

Summary
A simple comprehensive experimental approach has been provided to size emergency relief
devices in reactive service. Once credible reactive upset scenarios have been determined, low
phi factor calorimeters can be used to directly simulate the upset conditions and obtain data
regarding system classification, reaction rates, and flow regime. Using the simplified vent sizing
equation provided by Fauske, required vent sizes can then be determined from these data. The
applied methodology is consistent with available experimental data for all three types of reactive
systems.

References
Burelbach, J.P., Advanced Reactive System Screening Tool (ARSST), Presented at the 28th
Annual North American Thermal Analysis Society (NATAS) Conference, Orlando, Florida,
October 4-6, 2000.

Askonas, C.F., Burelbach, J.P., and Leung, J.C., The Versatile VSP2: A Tool for Adiabatic
Thermal Analysis and Vent Sizing Applications, Presented at the 28th Annual North American
Thermal Analysis Society (NATAS) Conference, Orlando, Florida, October 4-6, 2000.

Fauske, H.K., Properly Size Vents for Nonreactive and Reactive Chemicals, Chemical
Engineering Progress, pp. 17- 29, February, 2000.

Fauske, H.K., Leung, J.C., Fisher, H.G., Thermal Runaway Reactions in a Low Thermal Inertia
Apparatus, Thermochimica Acta, pp. 13-29, 1986.

Fisher, H.G., Forrest, H.S., Grossel, S.S., Huff, J.E., Muller, A.R., Noronha, J.A., Shaw, D.A.,
Tilley, B.J., Emergency Relief System Design Using DIERS Technology: The Design Institute
for Emergency Relief Systems (DIERS) Project Manual, American Institute of Chemical
Engineers, New York, NY, 1992.