Академический Документы
Профессиональный Документы
Культура Документы
~ - _
_ _ _ _ ~ ~ ~ _ -~
( DESIGNCASE \
T
+
-
I IDENTIFICATION
ACCEPTABLE
DETAlL/ PROCESS
1 I
ACCEPTABLE
ACCEPTABLE N
--
--
MITIGATION
ICoNTAlNMCNrl
I
WORST CREDIBLE
SCENARIO DESIGN BASIS <-
FIGURE 1. A strategy for emergency relief system FIGURE 2. An approach to emergency relief system
design. design assessment.
2 January, 1991 PlantlOperations Progress (Vol. 10, No. 1)
prohibitive, a three-step iterative approach can be used to arrive DIERS/DIERS Users Group
at an acceptable risk at minimum cost through optimization
of the measures taken t o prevent, moderate (relieve) and con- The Design Institute for Emergency Relief System (DIERS),
tain a runaway reaction or decomposition (Figure 2 ) . a consortium of 29 companies under the auspices of AIChE,
Many factors must be considered in arriving at the best was formed in 1976 to develop methods for the design of
approach to deal with hazards that may accompany runaway emergency relief systems to handle runaway reactions. Of par-
reactions. Thermochemistry, reaction kinetics, thermal sta- ticular interest were the prediction of when two-phase flow
bility, process conditions/controls, abnormal operation, con- venting would occur and the applicability of various two-phase
taminants, equipment design, equipment and instrument vapor-liquid flashing flow methods for sizing relief systems
failures, operating procedures and human error are all ex- [ 2 ] . DIERS spent approximately $1.6MM to investigate the
amined when evaluating hazard potentials. A n approach to two-phase vapor-liquid onset/disengagement dynamics and
the safe design of chemical processes which have the potential hydrodynamics of emergency relief systems. The DIERS pro-
for a runaway reaction is to identify and analyze worst credible gram has come t o a successful conclusion with publication of
incident scenarios by proceeding as follows: the research results [3-91 and preparation of a project manual
Make an exhaustive search for hazardous conditions by [lo].
involving safety specialists, reaction system designers, Over 60 companies have formed a DIERS Users Group to
process engineers and operating personnel. cooperatively assimilate, implement, maintain and upgrade the
Identify the sequences of events which could produce the DIERS methodology. The purpose of the new group is to:
highest pressure within a vessel and maximum flow from Reduce the frequency, severity and consequences of pres-
the emergency relief device@). sure producing accidents.
0 Scrutinize various failure modes t o arrive at the combi- Promote the development of new techniques which should
nation which produces the worst credible incident scen- improve the design of emergency relief systems.
ario. Membership is open to industrial or engineering organizations
0 Then, utilize reaction, control, process and safety engi- interested in the design, use or manufacture of emergency relief
neering design technologies to prevent, moderate (relieve) systems or devices [ I l l .
and contain runahray reactions.
Design and operating strategies which will help to prevent Two-Phase Vapor-Liquid Flow Onset and Disengage-
runaway reactions include: ment
Acquire data to identify potential problems. Emergency relief system design is a multifaceted problem.
Measure and control critical parameters (temperature, Of particular significance is whether the relief system must be
pressure, feed rate, coolant flow, catalyst level). designed for single phase (vapor or liquid) or two-phase vapor-
Operate at conditions (temperature, pressure, concentra- liquid flow. DIERS developed and tested a method to calculate
tion) which provide a safe margin from runaway condi- the onset (start) and disengagement (stop) of two-phase vapor-
tions. liquid flow from a vessel due to emergency relief or depres-
Jnstall redundant instrumentation t o increase reliability surization [I21 (See Appendix A).
of measurement and control of critical parameters.
Use alarms to warn operators that a critical parameter has
changed from its normal condition. Two-Phase Vapor-Liquid Flow Through Relief Devices
Train operators to safely react to upset conditions.
Identify never exceed limits. DIERS examined various two-phase vapor-liquid flow
Activate an emergency shutdown when a critical param- models from the open literature [4] and tested them using an
eter has deviated from normal by a predetermined amount overall system model against large-scale experimental data [ 5 ] .
or reaches a never exceed limit.
Prevent contamination by proper design and operating
Fauske 1131 provided practical guidelines for use and Huff [a
included many personal insights in his discussion of the various
procedures. two-phase flow models. Two-phase vapor-liquid homogeneous
These and many other steps ensure that a runaway reaction equilibrium flashing flow proved to be the most conservative
will not occur as the result of a single failure. model for estimates of flow capacity from both safety valves
Finally, the analysis of the likelihood and consequences of and rupture disks. However, this model is not necessarily con-
multiple failures leads to the identification of the worst credible servative for safety valve back pressure calculations or effluent
runaway reaction scenario. An emergency relief system can containment considerations because under certain circum-
then be designed to handle this scenario to include safe disposal stances the appropriate application of other models will predict
of the discharged fluid. In addition t o moderation of a runaway higher flow rates.
reaction by proper sizing of emergency relief, consideration Following completion of the DIERS program, Leung, et al.,
should also be given to installation of a liquid dump system, developed a generalized correlation for two-phase, homoge-
provision for emergency blowdown of pressure or use of a neous flow. These methods describe flashing, non-flashing
“kill” agent. (frozen) and mixed flashing/non-flashing (hybrid) choked and
Containment can be approached in two ways. First, vessels subcritical turbulent and laminar flow of single component
can be designed to withstarid the maximum pressure that can and multi-component mixtures through ideal nozzles (safety
develop from an upset. Although this approach may be viable valves) and horizontal and vertical pipes (rupture disks) [I#-
for some emergencies such as a vapor phase deflagration, it 221. The various modifications to the basic methodology are
may not be a feasible alternative for a runaway reaction or required t o design reliable emergency relief and containment
vessel fire exposure because of the extremely high pressure that systems.
can be produced.
Second, the term containment may also be used to describe
the disposaVdecontamination of the discharge from a relief Experimental Data for Emergency Relief Systems Design
system. Vent stacks, vapor-liquid separators, quench tanks,
scrubbers, flares, incinerators or combinations thereof can be A careful experimental program which uses representative
used to disperse, quench, :scrub, detoxify or burn the dis- samples is required to obtain data needed as a basis for emer-
charged fluid. gency relief system design. The present state of experimental
during vapor or two-phase flow venting is sufficient to NOTE: The ERS device is not able to control T(dT/dt >> 0)
control dT/dt at the set pressure of the ERS device during and P(dP/dt >> 0) at the set pressure.
the entire course of the reaction (i.e,, dT/dt = 0 or in- Gussy/Non-Tempered: Non-condensible gases are pro-
creases due only to a change of volatility). duced as a result of a decomposition reaction. The reac-
HybricUTernpered Reaction: Non-condensible gases are tants, products and/or solvent are not volatile or have an
produced as a result of a decomposition reaction. The extremely low volatility. The heat of vaporization during
reactants, products and/or solvent must be volatile. Heat vapor or two-phase flow venting is insufficient to temper
of vaporization cooling during vapor or two-phase flow the reaction at any point.
venting is sufficient to control dT/dt at the set pressure
of the ERS device during the entire course of the reaction. Emergency Relief System Design Techniques
NOTE: The ERS device is able to control T(dT/dt = 0) and
P(dP/dt = 0) at the set pressure. Techniques for sizing an emergency relief system for run-
Hybrid/Non- Tempered Reaction: Non-condensible gases away reaction include:
are produced as a result of a decomposition reaction. The Graphical or analytical design methods.
reactants, products and/or solvent, if volatile, are present Direct scaling of experimental data obtained in vessels
in insufficient quantity or have insufficient heat of va- with a very low thermal inertia.
porization during vapor or two-phase flow venting to tem- Computer simulation of incidents and flow through relief
per the reaction throughout its entire course. systems.
z
?
The technique selected depends upon the: NOTE: Thermal inertia increases substantially as a test vessel
Type and number of chemicals involved. empties during relief area to vessel charge scaling ex-
Availability of required process and experimental data. periments. Thermal inertia changes can have a signif-
Constraints imposed upon the designer. icant effect on the heat rate in the small vessel and the
scaled relief area for a large vessel.
A four-inch (100 mm) relaxation length is required to
Emergency Relief Sy,ytem Design-Analytical Methods attain homogeneous equilibrium flashing flow. Use of a
short nozzle or orifice in a small test vessel to simulate
Huff [36] has presented details of a pseudo-steady-state flow from a large emergency relief device is not conserv-
method, which makes use of closed system ARC data, for ative since a given homogeneous non-equilibrium flow is
design of emergency relief devices for volatile/tempered re- handled by a smaller area than an equivalent homogeneous
actions. Fauske has published a vent sizing nomograph [37- equilibrium flashing flow [28].
391 and analytical design methods [38, 391 for volatile/tem- The onset of two-phase vapor-liquid flow during a relief
pered reactions. Leung, el al., have discussed analytical design incident is dependent upon vessel height, except for a fluid
methods and closed form approximations for runaway reac- which exhibits a homogeneous vapor-liquid onset/dis-
tions with homogeneous [3,40-421 and churn-turbulent/bub- engagement regime. Materials which exhibit homogeneous
bly vessel behavior [43, 441. These methods include design (foamy/highly viscous) behavior will be essentially com-
approaches for volatile/cempered [40-421, hybrid/tempered, pletely emptied from either a small or full-size vessel. High
hybrid/non-tempered and gassy/non-tempered [41] reactions. vapor slip for a material with a partial (non-foamy/non-
Recently, FA1 published alternative analytical design methods viscous) onset/disengagement behavior may result in all
for each type reaction [29-311. The DIERS Bench-Scale Ap- vapor or limited two-phase vapor-liquid flow from a small
paratus can be used to provide required data for all of the test vessel while a much more extensive two-phase vapor-
emergency relief system design methods shown in Table 3 . liquid mixture can flow from a full-size vessel with the
same heat release per unit charge. Early disengagement
and potential runaway reaction rate turnaround in the
Emergency Relief System Design-Area: Charge Scaling small vessel will yield a non-conservative emergency relief
device area for a full-size vessel [27.
The DIERS Bench-Scale Apparatus can be used for direct
sizing of emergency relief devices by conducting top or bottom Runaway Reaction Emergency Relief System Design
vented tests (See Table 3). Relief area to vessel charge scaling Computer Programs
using top vented tests is valid for sizing top mounted emergency
relief devices for: Computer programs may be preferred or even required in
Volatile/tempered reactions which exhibit vapor or ho- some situations to correctly size safety valves, rupture disks
mogeneous disengagement behavior [27, 28, 461. or breather vents for process equipment. Since the underlying
NOTE: Area: Charge Scaling is not valid for fluids which science continues to be refined and advanced, these computer
exhibit partial disengagement (churn-turbulent or programs should contain the latest technology. This is the
bubbly) regimes. continuing mission of the AIChE Design Institute for Emer-
Hybrid/non-tempered and gassylnon-tempered reactions gency Relief Systems (DIERS).
which have a very high superficial vapor velocity and The non-linear heat and mass balance differential equations
therefore vent material more effectively and at a lower describing the transients of a worst credible runaway reaction
temperature from a large vessel [ 3 7 . incident and the fluid dynamic equations describing the flow
NOTE: Area: Charge Scaling is only valid if the test cell emp- capacity of an emergency relief system are normally solved by
ties. numerical integration in a digital simulation computer pro-
For volatile/tempered reactions which exhibit churn-turbulent, gram. The kinetics, stoichiometry, heats of reaction, physical
bubbly or homogeneous disengagement: properties and vapor-liquid equilibrium constants for the ma-
Top mounted emergency relief devices can be sized using terials present can be read into the computer program from a
bottom vented tests and a scaling equation [28]. data file. All physical properties should be calculated as a
Bottom mounted emergency relief devices can be sized by function of temperature in accordance with recognized ther-
relief area to vessel charge scaling using bottom vented modynamic models.
tests [ 4 4 . Program input and output describing the physical situation
NOTE: Sizing for hybrid or gassy reactions using bottom vent- should be checked for consistency and printed to provide a
ing is not valid because gas is lost through a pressure written record of the calculations. The mass of each component
equalization hole in the test cell. should also be continually checked to ensure a balance. A
variable step size in temperature and pressure may be used to
maintain numerical accuracy for stiff (rapid runaway reaction)
Limitations on Area: Charge Scaling for Emergency systems. Warning messages are usually provided if the emer-
Relief System Design gency relief devices cannot maintain the system pressure to
meet ASME pressure vessel code requirements.
Direct use of data to size an emergency relief device for a full- Huff [47 appears to have been the first to publish acomputer
size vessel by scaling on the basis of relief area to vessel charge simulation method to size emergency relief devices. His model
from a small, improperly designed vessel can lead t o a non- introduced the ‘‘uniform froth” or “homogeneous” concept
conservative design. Several factors can limit the usefulness of to allow simulation of two-phase venting. A slip equilibrium
some test vessels: hydrodynamics model for relief system flow was also docu-
The thermal inertia (ratio of the product of mass times mented.
heat capacity of the vessel plus sample to that of the sample Huff [471 subsequently refined and extended his simulation
alone) is high for most test vessels. Therefore, the ratio program t o cover the disengagement range from all vapor to
of the temperature rise of the sample plus heavy-walled all liquid relief. Additional two-phase hydrodynamic details
container is low compared to the temperature rise of the were provided.
sample alone. The experimental peak heating rate can Huff has summarized the basic method and equations in his
easily be reduced by a factor of one to two orders of versatile program and illustrated the use of the DIERS dis-
magnitude for a reaction with a high activation energy. engagement (churn-turbulent) and hydrodynamic (homo-
Best Estimate Procedure to Calculate Two-Phase Vapor- An example should serve to illustrate the phenomena. A4
Liquid Flow Unset/Disengagement two-inch diameter relief device (nozzle) was rapidly opened on
a tank which was 95 percent filled with 540 gallons of cit)
water at approximately 150°Cand under its own vapor pressure
(Ideal Prediction- Vertical Vessel) of about 58.5 psig. Approximately 28 percent of the tanh
contents vented by two-phase flow (Figure 3). The experiment
Overview
L
Liquid Swell
The DlEKS Bench-Scale Apparatus can be used to differ- where jgm - Superficial vapor velocity (ft/sec)
entiate between materials which exhibit homogeneous (foamy) F - Vapor flow rate (lb/hr)
versus non-foamy two-phase, vapor-liquid flow onset (start)/ pg - Vapor density (Ib/ft3)
disengagement (stop) behavior during runaway reaction emer- A, - Vessel cross-sectional area (ft2)
gency relief by measurement of the final void fraction in a test 3. Calculate the bubble rise velocity.
cell. Once data indicate 1hat a mixture is non-foamy ,the method
described below can be used to calculate ideal two-phase flow K[32.174 (2.2046E-3) u ( ~ f - p ~ ) ] ” ~
onset and disengagement behavior depending upon the vis-
urn= Pi
1/2
0.9
0.8
LL
0.5
H
0
>
w
W
0.4
U
Ly
0.2
0.1
0
0.1 1 10 103
D I M E N S I O N L E S S S U P E R F I C I A L VAPOR V E L O C I T Y . 9
NOTE: Figure 5 illustrates the disengagement relation- 5. Dimensionless Superficial Vupor Velociry at Mihich Two-
ships. phase Flow Commences (Churn-turbuleni)
2cy 2 (0.2)
$=-- - = 0.57
Sample Calculation 1 -COG 1 - 1.5 (0.2)
4. Dimensionless Superficial Vapor Velocity Due to Flow @ & = rl. = 1.04, ol = 0.29 from the CT (C, =-
+bP = jgm/Um= 0.637/0.610 = 1.04 1.5) curve.