THE REACTIVE SYSTEM SCREENING TOOL (RSST):

AN INEXPENSIVE AND PRACTICAL APPROACH

TO SIZING EMERGENCY RELIEF SYSTEMS

H. K. Fauske
Fauske & Associates, Inc.
Burr Ridge, Illinois

SUMMARY

An inexpensive practical approach to sizing venting requirements for reactive

chemical systems is outlined. Based upon easy to obtain RSST data, the ap-
proach provides a consistent comparison with all available large-scale data in-
cluding vapor, hybrid and gassy reactions.

1. INTRODUCTION

Knowledge of chemical reaction rates, character, volume is 10 ml and the containment volume is 350
and energy release is essential to modern process ml. A key feature of the apparatus is its low effective
design. The Design Institute for Emergency Relief heat capacity relative to that of the sample whose
Systems (DIERS) program (Fisher, 1985), sponsored value, expressed as the capacity ratio, is approxi-
by 29 companies under the auspices of the AIChE mately 1.04 (i.e., quite adiabatic). This feature al-
and completed in 1985, provided the chemical proc- lows the measured data to be directly applied to proc-
ess industry with tools necessary to gather such data ess scale.
(Fauske & Leung, 1985). A primary purpose of the
effort was evaluation of emergency relief vent re-
quirements, including energy and gas release rates for
systems under upset conditions and the effects of
two-phase flow on the emergency discharge process.
An easy, inexpensive approach to the DIERS proce-
dure was later provided by the RSST (Creed and
Fauske, 1990 and Fauske, 1993). This paper pro-
vides an update of the RSST methodology with em-
phasis on benchmarking the methodology against all
available large-scale data including vapor, hybrid,
and gassy reactive systems.

2. RSST DESCRIPTION
The RSST (Figures 1 and 2) consists of a spheri-
cal glass reaction vessel and immersion heater (op-
tional), its surrounding jacket heater and insulation,
thermocouples and a pressure transducer, a stainless Figure 1. Reactive System Screening Tool (RSST) contain-
steel containment vessel that serves as both a pres- ment.
sure simulator and safety vessel, and, not shown, a
magnetic stirrer base, a control box containing the
heater power supply, temperature amplifiers, and a
data acquisition and control panel. The sample cell

Presented at the 1998 Process Safety Symposium, October 26-27, 1998, Sheraton Astrodome Hotel, Houston, Texas
 more) evaporates from the glass vessel and condenses
on the cold containment walls, thereby removing the
reaction energy and tempering the system, as would
occur in a vented reactor. So long as there is suffi-
cient tempering liquid before the reactants are ex-
hausted, and the vent is sized to pass the vapor as it is
produced, tempering will limit the pressure rise. The
principal parameter determining the vent size is the
rate of temperature at the relief set pressure.

Figure 2. Reactive System Screening Tool (RSST) test cell.

Typically, a 10 mλ sample is heated at a constant

rate and the sample self-heat rate is found as a func-
tion of sample temperature. The imposed linear ramp
rate can be varied from 0.1ºC/min. to ramp rates ap-
proaching those required to simulate fire exposure by
using the immersion heater option. The heater com-
pensates for losses, and the self-heating rate of an
exothermic system is adjusted for heater input. Re-
agents may be added to a sample during a test. Data
handling programs produce plots of pressure vs. tem- Figure 3. Illustration of vapor, hybrid and gassy reactive sys-
perature, temperature vs. time, pressure vs. time, and tems.
Arrhenius plots.
The RSST quickly and safely determines the po- A gassy system has noncondensable reaction or
tential for runaway reactions and measures the rates decomposition products which continually increase
of temperature and, in case of gassy reactions, pres- the containment pressure and do not effectively tem-
sure increases to allow reliable determinations of the per the system, whose temperature may escalate rap-
energy and gas release rates. A recent additional idly. The total pressure is equal to the gas pressure,
feature of the RSST allows flow regime characteriza- and the maximum rate of pressure rise determines the
tion, i.e., "foamy" versus "non-foamy" behavior, to vent size. Note that a vapor system may evolve into
be determined under actual runaway conditions a gassy one when tempering is lost by complete boi-
(Fauske, 1998). This information can be combined loff of solvent.
with analytical tools (summarized later) to assess A hybrid exhibits both tempering and gassy char-
reactor vent size requirements. acteristics. Gaseous decomposition occurs prior to
boiling, but the rate of reaction (gas production by
decomposition) is still tempered by vaporization.
3. SYSTEM CHARACTERIZATION
The total pressure in the reactor is the sum of the gas
AND VENT SIZING partial pressure and the vapor pressure, and both the
rates of temperature and pressure rise are needed to
Three types of reactive systems are distinguished determine the proper vent size.
for their venting and mitigation character (see Figure
3). A vaporizing, or tempered system contains com-
ponents (often a solvent) whose vapor-liquid equilib-
3.1 Vapor System Sizing
rium controls the system temperature and reaction
The principal quantity of interest is the reaction
rate for a fixed containment pressure. This occurs
when one or more components boil prior to potential self heat rate, T&s (K s-1) at the relief set pressure, Ps
gaseous decomposition, so heat from exothermic (Pa) and temperature, T2 (K). The T& value is pro-
s
reactions is removed by the latent heat of vaporiza- vided by the RSST and can be applied directly to
tion, and the system pressure is equal to the compo-
nent vapor pressure. In the RSST, one compound (or

Presented at the 1998 Process Safety Symposium, October 26-27, 1998, Sheraton Astrodome Hotel, Houston, Texas
 & (m3
assess the vapor volumetric generation rate, Q v

s-1) from

&
& = m cT
Q (1)
λ ρv
v

where m (kg) is the reactant mass, c (J kg-1 K-1) is the

specific heat, λ (J kg-1) is the latent heat of vaporiza-
tion and ρv (kg m-3) is the vapor density. Considering
vapor venting only, the required vent area, Av (M2)
can be estimated from

1/ 2
3 m c T&  R Ts 
Av ≈   (2) Figure 5. Illustration of "foamy" or "bubbly-like" experimen-
2 C D λ Ps  M w,v  tal data (ATP) and comparison with all vapor venting (Av).

1/ 2 The vent sizing formula, A (in2) = 0.053 V (gal),

m c T&  R Ts  based upon large-scale experience and used for phe-
Av ≈ 1/ 2   (3)
C D Ps  2 (1 − Pb / P )   M w,v  nol-formaldehyde reactors (Howard, 1973), is used to
  illustrate the "overpressure" effect on venting re-
quirements in Figure 5 for "foamy" or "bubbly-like"
for critical and highly subcritical flow conditions, systems. Noting that the Monsanto formula is based
respectively, where CD is the discharge coefficient, Pb upon a Ps value of 1-2 psig and a T&s value of about
(Pa) is the back pressure, R (8314 Pa-m3/K-kg-mole)
6.5ºC min-1, we estimate that the formula represents
is the gas constant, and Mw,v is the molecular weight.
an "overpressure" of about 17%. At this overpressure
Vent areas from available large-scale experiments
with "non-foamy" or "churn turbulent" like systems the value of ATP/Av is about 4.9∗ (see Figure 5). The
(10 m3 methanol/acetic anhydride, Linga et al., 1998, noted overpressure of 17% and the solid curve repre-
and 2 m3 high conversion (HC) ethylbenzene/styrene, senting the "overpressure" effect are obtained by
Fisher, et al., 1992) are compared to values from combining the Monsanto formula with the homoge-
Equation 2 in Figure 4. The solid lines through the neous vessel and vent flow formula ATP = (m T& )/2
normalized data illustrate the well-known "overpres- (Ts/c)1/2 ∆P, where ∆P (Pa) is the overpressure
sure" effect in reducing the vent area when two-phase (Fauske, 1989). The latter formula is in good agree-
flow is occurring (Leung, 1987 and Fauske, 1989). ment with the low conversion (LC) ethylben-
In fact, both theory and the data show that for "non- zene/styrene data (Fisher et al., 1992) exhibiting
foamy" systems and allowance of modest "overpres- "bubbly-like" behavior (see Figure 5). The interpre-
sure", the required vent area can be assessed consid- tation of the Monsanto formula is also in excellent
ering vapor venting only, i.e., Equation 2. agreement with recent 22λ bottom vented phenol-
formaldehyde tests (Leung et al., 1998). For a meas-
ured overpressure of 21.7% the value of ATP/Av is
about 3.8 (see Figure 5).
Considering the enveloping nature of the phenol-
formaldehyde system with its water-like properties,
both theory and large-scale experience suggest that
for "foamy" systems and allowance of modest "over-
pressure", an adequate vent size can be based upon
large-scale fire exposure experiments with "foamy"
and "non-foamy" water systems (Fauske, 1986).

3.2 Hybrid System Vent Sizing

Figure 4. Illustration of "non-foamy" experimental data (ATP) ∗

It is of interest to note that for a Ps value of 1-2 psig and zero
and comparison with all vapor venting (Av). overpressure, this ratio would be about 160 based upon flashing
two-phase flow condition.

Presented at the 1998 Process Safety Symposium, October 26-27, 1998, Sheraton Astrodome Hotel, Houston, Texas
 The unique design offered by the RSST is ideally
suited to characterize vent requirements for hybrid
systems. The measured self heat rate, T& (K s-1), and
rate of pressure rise, P& (Pa s-1), for a given relief set
pressure, Ps (Pa), can be applied directly to assess the
total gas-vapor volumetric rate, Q & (m3 s-1)
T

& &
& =Q
Q & = mcT + v P m
& +Q (4)
λ ρv
T v g
Ps m t

where v (3.5 • 10-4 m3) is the RSST containment vol-

ume and mt (kg) is the RSST sample mass. Consider- Figure 6. Comparison of H2O2 decomposition data with pre-
ing gas-vapor venting only, the vent area to volume diction from the RSST venting formula.
ratio, A/V (m-1) can be estimated from
at tempering of 70ºC min-1, would only result in
1/ 2
3  ρ c T& ρ v P&  M w  about 30% overpressure (∼ 4.4 psi) in case of exercis-
A/V ≈  +   (5) ing the DIERS methodology. The lesson learned
2 CD  ρ v λ m t Ps   R Ts  here, is that allowance for a modest overpressure
eliminates the need to consider uncertainties related
1/ 2
1  ρ c T& ρ v P&  Mw  to flow regimes and non-equilibrium effects.
A/V ≈  +   (6)
C D  ρ v λ m t Ps   2 (1-Pb / Ps ) R Ts 
3.3 Gassy System Vent Sizing
for critical and highly subcritical flow conditions, For these systems Equation 5 reduces to
respectively, where ρ (kg/m3) is the loading density.
The above vent sizing methodology for hybrid 1/ 2
systems is consistent with the large-scale 200 kg 50 3 ρ v P&  M w 
A/V ≈   (7)
wt% H2O2 runaway reaction trials reported by 2 CD m t P  R T 
Wilberforce (1988), as illustrated in Figure 6. The
absence of two-phase flows as the tempering is ap- and the maximum rate of pressure rise, P& , is of prin-
proached for these test trials is another example of cipal interest and again is provided by the RSST.
the flow regime complexity during runaway reaction Available venting data for peroxide systems, 37.5
conditions. Apparently in these trials the vapor-gas wt% 3,5,5 trimethyl hexanoyl peroxide (Wakker and
release is highly non-uniform due to inherent non- deGroot, 1996), neat dicumyl peroxide (Gove, 1996),
equilibrium conditions with the majority of the re- and t-butyl peroxy bensoate (Wakker and deGroot,
lease occurring at the free liquid surface, resulting in 1996) are compared to predictions from Equation 7 in
insignificant liquid swell in the absence of "foamy" Figure 7, illustrating good agreement with both low,
conditions (Fauske et al., 1986). intermediate and high peroxide energetics levels.
Considering the often used DIERS two-phase
The RSST measured peak P& values for these systems
flow methodology of "churn-turbulent" flow regime
with Co = 1.5 (Fisher et al., 1992), a self-heat rate at (mt ≈ 0.01 kg) are 1000, 4000 and 100,000 psi min-1
for 37.5 wt% 3,5,5 trimethyl hexanoyl peroxide, neat
tempering in excess of 3.4ºC min-1 would result in
dicumyl peroxide and t-butyl peroxy benzoate, re-
liquid ejection or two-phase flow, which is clearly
spectively. The predictions for neat dicumyl perox-
not the case. However, we note that consistent with
ide are also consistent with incident data reported by
the above discussion for vapor systems, a vent area
Gove (1996).
based upon all gas-vapor venting for a self heat rate

Presented at the 1998 Process Safety Symposium, October 26-27, 1998, Sheraton Astrodome Hotel, Houston, Texas
 Fauske, H. K. and Leung, J. C., 1985, "New Experi-
mental Techniques for Characterizing Runaway
Chemical Reactions," Chem. Eng. Prog., 81(8),
39-46, August.

Fauske, H. K. et al., 1986, "Emergency Relief Vent

Size for Fire Emergencies Involving Liquid-Filled
Atmospheric Storage Vessels," Plant/Operations
Progress, 5(4) 205-108, October.

Fauske, H. K., 1989, "Emergency Relief System De-

sign for Runaway Chemical Reaction: Extension
of the DIERS Methodology," Chem. Eng. Res.
Des., Vol. 67, 199-202, March.

Figure 7. Large-scale runaway peroxide decomposition data Fauske, H. K., 1993, "Reactive System Screening
and comparisons with RSST simulation with all gas venting.
Tool," United States Patent Number 5,229,074,
July 20.
We note that the DIERS' methodology for gassy
systems considering initiation of two-phase flow at
Fauske, H. K., 1998, "Flow Regime Characterization
the measured peak reactive conditions and no prior
During Emergency Relief," FAI Process Safety
material loss overestimate the vent areas noted in
News, Summer 1998, Vol. 5, No. 2.
Figure 7 by at least an order of magnitude. As indi-
cated by the RSST tests, significant material losses
Fauske, H. K., 1998a, "The Reactive System Screen-
occur well before reaching peak reactive conditions,
ing Tool (RSST): An East Inexpensive Approach
further justifying the assumption of gas venting only.
to the DIERS Procedure, Int. Symp. on Runaway
Furthermore, it is of interest to note the transition
Reaction, Pressure Relief Design, and Effluent
from a homogeneous to propagating reaction behav-
Handling, March 11-13, New Orleans, Louisiana.
ior as the peroxide energetics level increases (Fauske,
1998a and Grolmes, 1998). Such transitions strongly
Fisher, H. G., 1985, "DIERS Research Program on
influence the peak volumetric gas generation rate and
Emergency Relief Systems," Chem. Engr. Prog.,
require experimental determination as provided by
81(8), 33-36, August.
the RSST. The transition, as well as the rage of
propagation cannot be predicted by theoretical
Fisher, H. G. et al., 1992, "Emergency Relief System
means.
Design Using DIERS Technology," American In-
stitute of Chemical Engineers, New York, NY.
SUMMARY
Gove, S. H., 1996, "Emergency Pressure Relief for
Due to its ease of use as well as its low cost, the Intermediate Bulk Containers Containing Di-
RSST is now a standard industry tool for chemical cumyl Peroxide," DIERS User Group Meeting,
system characterization for relief system design data February 5, Houston, Texas.
acquisition. The RSST data which can be scaled di-
rectly to full-size applications have been demon- Grolmes, M. A. 1998, "Pressure Relief Requirements
strated to produce excellent agreement with a large for Organic Peroxide and Other Related Compo-
number of large-scale venting tests including vapor, nents," Int. Symp. on Runaway Reaction, Pressure
hybrid and gassy systems. Easy to use design equa- Relief Design, and Effluent Handling, March 11-
tions are provided for these systems. 13, New Orleans, Louisiana.

REFERENCES Howard, W. B., 1973, "Reactor Relief Systems for

Phenolic Resins," Monsanto Company, June.
Creed, M. J. and Fauske, H. K., 1990, "An Easy, In-
expensive Approach to the DIERS Procedure," Leung, J. C., 1987, "Overpressure During Emergency
Chemical Engineering Progress, pp. 45-49, Relief Venting in Bubbly and Churn-Turbulent
March. Flow," AIChE Journal, 33(6), 952.

Presented at the 1998 Process Safety Symposium, October 26-27, 1998, Sheraton Astrodome Hotel, Houston, Texas
Leung, J. C. et al., 1998, "Phenolic Runaway Reac-
tion - Pressure Relief and Containment," Int.
Symp. on Runaway Reaction, Pressure Relief De-
sign, and Effluent Handling, March 11-13, New
Orleans, Louisiana.

Linga, H. et al., 1998, "Large-Scale Runaway Reac-

tion Tests," Paper presented at the Loss Preven-
tion Symposium, AIChE Spring National Meet-
ing, March 8-12, New Orleans, Louisiana.

Wakker, J. P. and deGroot, J. J., 1996, "Venting of

Decompositions of Energetic Liquids Using Bot-
tom Vent," Proc. of Process Plant Safety Sympo-
sium, Vol. 2, pp. 55-71, Houston, Texas.

Wilberforce, J. K., 1988, "Emergency Venting of

Hydrogen Peroxide Tanks," CEFIC Hydrogen
Peroxide Safety Conference, Gotherburg, Sep-
tember 22.

Presented at the 1998 Process Safety Symposium, October 26-27, 1998, Sheraton Astrodome Hotel, Houston, Texas