REFINING
Crude column relief design
Examination of a crude unit's behaviour using a dynamic model can provide
significant insight into the operation and transient behaviour of the process,
preventing unnecessary investments in relief system upgrades
Uwe Nagel OMV Deutschland GmbH Howard B Jemison and Cal Depew SimSci-Esscor
Ralph-Uwe Dietrich Invensys Process Systems
MV Deutschland Gmb#t (MVD)
recently completed a revamp of
the crude distillation unit (CDU)
at the Burghausen reflnery. This revamp
Jnchided mechanical design modifica
tions that requlzed the re-ating of the
column's design condition and. the
implementation of a highly reliable,
redundant emergency shutdown (ESD)
system. German authorities required,
recertification to confirm that the
‘existing relief systems could handle
fapected relief scenarios and were in
compliance with safety regulations with
the new ESD system in place. SimSci-
‘sscor carried out an engineering study
to determine the maximum column
pressure and relief loads of the OMVD
fnide unit for various abnormal
conditions and. relief scenarios. The
study employed a rigorous dynamic
model ofthe crude tower and associated
squipment,
Rigorous dynamic process simulation
proved to be an effective supplement to
Conventional methods when these
produced unrealistically high rellet
Toads. A dynamic model generates far
‘more realistic, yet still conservative,
predictions of CDU relief loads. The
Geral and thermodynamic rigour of the
dynamic simulation model allow it to be
validated to actual plant dynamic
behaviour. The validated model can
then provide high-quality rellet load
predictions.
‘A pre-condition of the use of the
dynamic process model results for relief
studies was the approval ofits accuracy
by both the OMVD operation personnel
and the German authorities, The model
was validated against the results from
both the OMVD steady-state and
transient plant operating data. The
model reached excellent agreement (<1
ditference) with the OMVD steady-state
crude unit flows, temperatures and
pressures. A comparison with the
ldynamie plant operating data showed a
good match between plant -measure-
ments and model results. Some
snismatches in pressure and temperatures
were identified, but the model was
designed to have slightly moze and
lighter material in the overhead system
to provide a more conservative estimate
cof the relief loading during abnormal
‘conditions. The model also had a sight
faster dynamic response on the upper
temperatures than the actual plant, but
this difference will provide 2 more
pessimistic estimate of column,
Dehaviout. With these discrepancies
explained, it was determined that the
dynamic model was properly validated
and accurate for the pressure relief stud.
‘New ESD logic was included in the
dynamic model, as well as all the
necessary regulatory control for normal
plant operation, ESD logic is a certified
redundant system approved by the TOV.
Siiddeutschland (TOV) as failsafe. The
regulatory controls that improved the
safety and stability of the crude unit
‘vere not inchided in the model (ie, the
overhead temperature controller of the
‘overhead pressure controller that vents
to the flare), but the controls that may
Increase the reli load were included.
The dynamic model test cases
Indicated ‘that the current shutdown,
logic and pressure relief systems. were
capable of handling all of the
investigated relief scenarios. These
conclusions enhanced the results of
carlier calculations using conventional
mcthods, and the dynamic study results
‘obtained approval by the TOV. No
additional equipment for the relief
“a dynamic simulation study
of a process can provide a
more accurate and
quantitative understanding
31
Pra az 2008
wwwepta.com
system was recommended and wnneces-
sary relief system upgrade investment
wvas prevented,
“The proprietary Dynsim model
provides Very reliable indications of the
dynamic responses of the crude unit to
various abnormal conditions and failures
inorder to predict the possible loading on
the relief system, This study strongly
confirms that the use of dynamic
Simulation technology complements
conventional calculation techniques by
providing more accurate models that
Consider the transient behaviour of the
process and control system. Further use of
the model to improve the unit's control
regime or as a basis for an operator
‘taining simulator system is possible.
Dynamic simulation
Refineries continuously debottleneck
key processing units such as CDUs for
improved separation to stay competitive
and increase throughput. Following this
path, the predicted column relief load
(estimated using eanventional methods)
eventually reaches the capacity of the
Connected relief system. Therefore,
Significant non-prodietive investments
say be necessary to comply with acept
cd safety standards and regulations.
In recent years, the use of dynamic
simulation has. become increasingly
common inthe design and specification
ff pressure relief systems. dynamic
Simulation study of a process can
provide a more accurate and quantita-
tive understanding of a pressure zlit
event, including information not
avallable via conventional techniques.
Before relevant conclusions canbe
obtained from a simulation, the
‘dynamic simulation madet should be
validated against operating data. to
prove its accuracy, if possible
Tm this example, the crude unit
revamp included the implementation of
new, enhanced ESD logic. A mechanical
resting of the column identified that
the section of crude tower where the
liameter is reduced (the cone section)
required a lower allowable workingREFINING
Profash 1 Crude cetiation
Mower
Figure 1 Schematic diagram of the OMVD crude tower and all necessary associated
‘equipment for on accurate representation of the refinery’s crude distillation process
E Motes eqipment
= certs
ee, uncer
sie cle Teh poe
creator fame) Roe ceane
Sane ron eas ga
matte conocer ie 0503)
Table 1
pressure for the column. Therefore, the
refinery looked for recertification that
‘the existing relief system was stil able to
handle expected tellef scenarios and
that the relief would be fast enough to
maintain the crude tower pressures
vithin limits. Hopefully, the revamp
would not require additional invest-
sents in the reli system, The primary
goal of the study was to reassure the
‘operating staff and the safety authorities
that the existing relief system was
adequate for the tested emergency cases
Inadaition, the study would provide the
engineers and operators with a better
‘understanding of the dynamic response
of the crude unit during abnormal
situations.
If crude refiners reviewed thetr plant
history, very few cases of crude unit
rellef valve openings would be observed,
fand any such events result in almost
immediate column pressure reduction
Maximum column pressures (design
pressure plus allowable overpressure) are
rarely reached due to conservatively
sized relief and flare systems, The
challenge is to apply this experience to
an appropriate CDU relief system
design, while ensuring that operations
meet the safety and enviconmental
protection requirements. Since the mid-
1970s, considerable development effort
has been spent by industry groups, such
as the American Institute of Chemical
Engineers (AIChE) design institute for
emergency relief systems (DIERS), to
improve methods for reliet flow
prediction and to standardise pressure
relief device sizing methodologies
Conventional methodologies make
many assumptions concerning the
nature of the event, which often lead to
total pressure relief loads that are overly
‘conservative, thas predicting relief loads
‘that are in excess of the actual relief
loads experienced in the refinery. These
methods da not account for the
following conditions
‘= Strong Interaction between heat.
Integrated equipment, such as. pump:
around heat exchangers
33
Pra 02 2006
— Interactions between different relief
devices connected 10 a common flare
header system
= Time delay to reach the relief
pressure after an event, and the
‘maximum pressure obtained
"Change of relief flow composition
iter the intial opening of the valve
— The effect of operating controls is
not considered, although some of the
control responses will happen automat
leally (le, emergency shutdown Logie),
some regulatory control action may
exacerbate a relief scenario,
In recent years, the use of dynamic
simulation to predict more accurate relict
leads has become “used and accepted” in
the petroleum and petrochemical
Industries" Some literature sources
conclude that the overestimation of
total relief rates and loads based on these
procedures is. S0-200%, An accurate
rigorous dynamic model of the process,
‘major contol loops and the associated
Telief system (including the pressure
relle valves) provides a more detalled
and accurate description of the process
Interactions during a relieving scenario.
Work scope
“The study employed a rigorous dynamic
model of the crude tower and associated
equipment using the previously noted
Dynsim software. The model scope was
discussed in great detail 10 include
everything that might influence the
relict flows and pressures using previous
refinery operation experience. The final
‘model scope is summarised in Figure 1
and a lst of major equipment included
in the model is given in Table 1. All the
equipment provides accurate estimates
fof mass_and energy balances and
rigorous thermodynamic calculations
For example, the individual tray
geometry in the erude tower is inched
in detall to ensure comet calculations of
hydraulic effects and hold-ups. Each
tray is modelled rigorously, taking ito
account the following:
= "Calculations for variable liquid hold-
= Tray hydraulic equations that relate
liquid flow off the tray and the height of
liguid over the weir
—Non-equilibrium flash calculations
based on tay efficiency
— Rigorous pressure drop calculation
based on the vapour flow rate and liquid
level on the tay.
The process heat exchangers were
modelled using the rigorous mti-phase
‘models that use the rating equations as
the defining equation for heat transfer
‘The vessel equipment provided models
forthe reflux accumulator and naphtha
accumulator, including the following
functionality
— Total volume and level calculationsREFINING
based on the vessel's length, diameter Model validation is required for trusting designed to insure that the dynamic
and orientation the results and predictions of any results fom the model were consistent
sndvjuse separation of feeds into process model. As previously mentioned, with actual plant behaviour dealy, the
Tapote and Ie holdsups within the Operators sequire model accuracy and dynamic model should be, valldated
Taper using a agorous Mash calculation, reliablty at a very high level, especially against actual plant data that js similar
Min en optional water phase decanted i they expect fo get relevant safety to the actual operating conditions of
fom the liquid phase in a water boot information. interest. However, for plant relief studies,
froniter Tphise calculations, if Two types of valldation tests were it is impractical and unwise to expose
appropriate performed, Fist, the steady-state results. the plant to a tellef scenario jut to get
aprile. pressure calculation based from the dynamic model were dynamic plant test data for validation of
gm the vapour hold-up and volume compared to several sets of steady-state the dynamic model. Thus, a more
inside the vesse] plant data and accepted run results with practical and safer plant test was applied
insifiguld vapourisation calculations the proprietary PROMI software. Pure For this dynamic valkation test =
hen appropriate (lor example, when and pseudo-component data, feed step change wes made In the reflux flow
the cellef valves open). Eompositions and initial estimates of rate by about 15%. The dynamic
‘the tap locations and inlet and exit intermediate temperatures, pressures responses of Key measurements in the
locations provide accurate calculations, and compositions coukl be taken from overhead section of the crude tower
cate Cro Vie vessel dries out or the PRO/II Tun, The dynamic model was were then recorded. Careful test
‘becomes liquid-full fumed by adjusting tray efficiencies, heat planning and operations execution ae
care aaitient and control logic transfer coefficients and pressure drops ecessary to keep all other process
database for the process model of the to match the steady-state plant data, conditions ‘under control during the
cea uier and its associated equip. Excellent agreement was obtained brief test period. Crude oll feed rate and
caine oer maiucted using the control between the plant data and the model: composition should remain constant to
ret Towle modules available in Dynsim. — All iynsim model product flow rates get comparable ‘dynamic plant, data
The model used over 20 controllers, $0. matched the flow measurements Using these data, the model was
camel and a variety of logic — Minimal pressure discrepancies adjusted using the same tuning
modules Tasted Between model and plant parameters as before to match the
reasurements Gynamic plant data, The tuning
Model validation miecThe product temperatures as well constants from the DCS controls were
aroceronced process engineer must the intemal column temperature profile Incorporated into the analogous control,
Re oPerGynamie model fesuls with were met within