You are on page 1of 7

Fluids/Solids Handling

Size Letdown
Lines for Pressure Surges
Loss of liquid in high-pressure
liquid/vapor separators may
result in a high-pressure gas flow in
the letdown circuit, causing damage
to piping and equipment. Here is
how to design the system to avoid
Gianni Anci, damage should a pressurized gas
EPSI
flow occur.

M any petroleum refinery processes (e.g., distillate hydrotreating,


gas sweetening) as well as petrochemical processes (e.g.,
methanol synthesis) employ liquid/vapor separators operating at
high pressures. The loss of the liquid level in such equipment
may cause sudden injection of high-pressure gas into the letdown circuit down-
stream of the control valve.
Under such conditions, the gas expands at the valve exit, but due to the limited
volume available, a pressure buildup occurs before the liquid is pushed away by the
gas. The inertia of the liquid and its delay in increasing its velocity, which limit the
volume available to the gas for expansion, are likely to cause a sudden pressure
surge in the piping circuit immediately downstream of the control valve. The prob-
lem faced by the design engineer is to evaluate the extent of the pressure rise, and
to establish adequate mechanical design conditions for the various elements of the
circuit. A relief device may be required to limit the maximum pressure rise during
transient conditions.
We studied this problem and performed simulations as a means to develop a
safe design basis. The study assessed the main consequences of the injection of
high-pressure gas into the letdown circuit, and looked at the fluid dynamics in-
volved, such that safety provisions would be adequate.
The study results indicated:
• Immediately upon loss of level, gas is injected in the low-pressure letdown
circuit, causing a sudden pressure rise, and consequently a water-hammer-type
transient.
• A pressure wave propagates along the line from the control valve to the
first downstream vessel, pressurizing the whole circuit.
• The liquid is displaced at a high velocity by the gas. Consequently, strong
acceleration forces involving severe thrusts on supports develop at any change
of direction.
• The stresses due to these acceleration forces are added to those from the pres-
sure increase, and may cause tube-rupture, if not considered in the design.
• The pressure rise is more pronounced for systems in which only a small vol-
ume of gas is released by the liquid flashing in the control valve. A large volume of

42 www.cepmagazine.org November 2001 CEP


gas attenuates the surge and reduces the speed of propagation can effectively avoid some of the potential risks. The follow-
of the pressure wave. ing precautions are recommended:
A calculation method is now presented for the design. 1. Preferably use the same rating and design conditions
This method offers a reasonably conservative estimate of the for the letdown line downstream of the control valve as for
pressure rise during the transient, and permits a safe defini- the line upstream, connecting the valve with the high-
tion of the required maximum allowable working pressure pressure separator.
(MAWP) for the piping. 2. Locate the control valve as close as possible to the
expansion vessel. This minimizes the length of the letdown
Design recommendations line downstream of the valve, and consequently reduces the
The results of the simulations indicate the need to consid- mass of product that suffers the acceleration and develops
er the phenomena in the hydraulic and structural design of the pressure surge.
the letdown circuit, and suggest a number of provisions that 3. Limit oversizing of the control valve to minimum
practical requirements.
4. Consider having the manual bypass to the control valve
Nomenclature closed and sealed if a bypass is installed.
5. If any equipment is installed in the downstream circuit
a = speed of the pressure wave in the liquid, m/s and has a MAWP lower than the upstream separator’s operat-
CV, liq = liquid-service valve flow-coefficient, dimensionless
ing pressure (or lower than the estimated peak-surge pres-
CV, gas = gas-service valve flow-coefficient, dimensionless
D = internal pipe dia., m
sure), consider installing a quick-acting safety relief device
E = Young’s modulus of the pipe material, kg/m2 (e.g., a rupture disk).
f = friction factor, dimensionless 6. The relief device should be installed preferably imme-
Fk = gas-specific-heat ratio factor (air = 1), dimensionless diately downstream of the control valve, but, in any case, up-
FP = piping geometry factor (reducer correction), dimensionless stream of the equipment to be protected.
g = acceleration of gravity, m/s2 7. Take special care in designing the supports of the
H = piezometric head, m downstream line, with due consideration to the thrust oc-
k = ratio of specific heats, dimensionless
curring at any change of direction, as with elbows or tees
Keff = effective bulk modulus of fluid, kgf/m2
Kgas = bulk modulus of the gas phase, kgf/m2
that are connected to the high-velocity fluid line or in the
Kliq = bulk modulus of the liquid phase, kgf/m2 route of the gas/liquid interface (associated with a sudden
L = pipe total length, m density change).
M = gas molecular weight
Pa = pressure assumed for first secant pole, bar Method of analysis
Pb = pressure assumed for second secant pole, bar The model is based on a method normally used for water-
Pc = pressure causing critical flow, bar hammer pressure-surge calculations, adapted to account for
P1 = upstream pressure, bar
the boundary conditions corresponding to the gas injection in
P2 = downstream pressure, bar
Psep = pressure in upstream separator, bar
the upstream section when level loss occurs in the separator.
∆P = pressure drop, bar A comprehensive general treatment of water hammer and re-
s = pipe thickness, m lated problems is presented in Wylie et al. (1).
t = time, s A number of simplifying assumptions were made in de-
T1 = absolute inlet temperature, K veloping the model to avoid excessive complexity and com-
V = liquid velocity in the pipe, m/s putational difficulty, yet maintain sufficient accuracy in the
W = mass flowrate, kg/h calculated values of pressure buildup, fluid velocity and ac-
x = distance along the pipe, m
celeration — all required for a sound mechanical design of
Xp = pressure drop ratio, ∆P/∆P1
Y = gas expansion factor, dimensionless
the pipeline and relevant supports.
Z = compressibility factor, dimensionless
XT = pressure drop ratio factor, dimensionless Assumptions
The model assumes that:
Subscripts • The loss of liquid level and the injection of gas in the
A, B. P = points in Figure 1 pipeline downstream of the control valve occur starting from
steady-state conditions, corresponding to the operating liquid
Greek letters
flowrate (for a conservative design, use the maximum operat-
γ = specific gravity of liquid, dimensionless
ε = volumetric fraction of gas in fluid, dimensionless
ing liquid flowrate in the circuit).
λ = dimensionless multiplier for combining Eqs. 4 and 5 • The fluid motion in the pipeline is considered one-
ρgas = gas density, kg/m3 dimensional; space integration is performed only along the
ρliq = liquid density, kg/m3 x-axis.
ρmix = mixture average density, kg/m3 • The fluid in the pipeline is homogeneous. In reality, de-

CEP November 2001 www.cepmagazine.org 43


Fluids/Solids Handling

pending upon the type of fluid and the operating pressures, Fluid properties
flashing can occur in the liquid being let down from high to The model uses average properties for the fluid (liquid
low pressure. The presence of gas bubbles or even a second containing dispersed gas):
phase reduces the apparent overall or “effective” value of the The mean fluid density is evaluated with Eq. 1:
bulk modulus and the speed of propagation of the pressure
wave in the fluid, thus reducing the severity of the pressure ρmix = ε ρgas + (1 - ε) ρliq (1)
surge for single-phase conditions. This is accounted for by
adopting suitably corrected bulk properties for the fluid. This The speed of propagation of sound in the fluid, is evaluat-
approach was experimentally verified to provide accurate re- ed by using:
sults for volumetric fractions of gases in the liquid up to 3%
(2). a= K eff /ρ mix (2)
• The fluid injected from the upstream separator during
the transient consists of gas only: entrained liquid, if any, is and is influenced by the fraction of gas in the gas/liquid mix-
considered to contribute a negligible fraction to the total vol- ture, and by the elastic properties of the pipe. An effective
ume upstream of the gas/liquid interface. bulk modulus is defined by:
• The inertia effect and the friction losses of the gas are
considered negligible; the pressure in the volume (pipe seg- Keff = 1/[(1 - ε)/Kliq + ε/Kgas + d/Es] (3)
ment) occupied by the gas is considered uniform.
• The gas pressure follows the ideal gas law. At low temperatures and moderate pressures, the gas-
• Neither heat nor mass transfer occurs at the gas/liquid phase bulk modulus may be approximated by the local gas
interface. absolute pressure.
• The pressure upstream of the valve in the gas/liquid sep-
arator remains constant throughout the computation period. Basic equations
• The gas injection rate is calculated for a constant open- The basic equations in water-hammer analysis are the
ing of the control valve (i.e., assuming the valve coefficient equations of motion and continuity. By considering an ele-
corresponds to steady-state operation at liquid design rate). ment of fluid in a pipe, the equation of motion may be ex-
• The compression of the gas in the gas-filled portion of pressed by Eq. 4, while the equation of continuity may be ex-
the pipeline is isothermal. pressed by Eq. 5:
The model considers the injection of gas into the pipeline
fV V
through the control valve, with the gas occupying the vol- g ∂H + V ∂V + ∂V + =0 (4)
ume of pipe between the valve and the previous fluid inter- ∂x ∂x ∂t 2D
face (possibly a liquid with dispersed gas). The gas/fluid in-
terface moves along the pipe, depending on the initial a 2 ∂V + ∂H + V ∂H = 0 (5)
(steady-state) conditions of the liquid, and on the evolution g ∂x ∂t ∂x
of the pressure at the interface. The pressure buildup in the
line accelerates the fluid, which offers resistance due to iner- Equations 4 and 5 contain the two unknowns V and H. We
tia and friction. can consider a linear combination of the two equations, in the
The pressure on the gas/fluid interface is calculated using form of Eq. 6, that can be rearranged in the form of Eq. 7.
the valve’s flowrate equation and mass balance, applying the
perfect gas law; it converges on the equilibrium pressure by Eq. 4 + λ Eq. 5 = 0 (6)
using the secant method. The transient pressure and velocity
variations in the fluid are calculated by the equation of mo- λ is a dimensionless multiplier used for the linear combi-
tion and the equation of continuity, following the classical nation of the two equations. The linear combination permits
method used in water-hammer analysis. transforming the partial-derivative equations into total-deriva-
The integration of the differential equations is performed tive ones, and enables solving them under some conditions.
by the method of characteristics, with finite-difference ap-
proximations in time and space. Piping components are ∂H V + λg + ∂H + λ ∂V V + a 2 + ∂V +
characterized by mathematically formulating appropriate ∂x ∂t ∂x gλ ∂t
boundary conditions describing the corresponding flow
properties. For simplicity, no inline valves or other compo- fV V
λ =0 (7)
nents are considered, as a simple pipe appears to be general- 2D
ly adequate and sufficient to characterize most letdown cir-
cuits. If more complex components or equipment are pre- If Eq. 8 and 9 are satisfied, then the first bracket of Eq. 7
sent, specific boundary conditions or mathematical models would be the total derivative, dH/dt, and the second bracket
should be adopted. would be the total derivative, dV/dt, as in Eqs. 10 and 11:

44 www.cepmagazine.org November 2001 CEP


∆t ∆x ∆x ∆x ∆x ∆x
∆x
P
∆t

∆t ∆t
P

∆t C+ C-

A C B
x
A R C S B
■ Figure 1. Characteristic curves in space-time region.
a ∆t
RC = AC *
∆x
■ Figure 2. Interpolation scheme used in solving equations.
dx = V + λg (8)
dt
and C-, at Point P, the values of the unknowns must satisfy
dx = V + a 2 both equations. So, the two equations may be solved for the
dt λg (9)
two unknowns to yield the values of head, H, and velocity, V.
At this point Eqs. 14 and 16 may be solved for x and t.
dH = ∂H dx + ∂H Consequently, the solution is carried out along the character-
dt ∂x dt ∂t (10)
istics, starting from known conditions, by finding new inter-
sections, so that velocity and head values are calculated for
dV = ∂V dx + ∂V the next time-step.
dt ∂x dt ∂t (11)
Finite differences
Equations 8 and 9 must be equivalent, which implies that For the purpose of calculation, the pipe is considered to be
Eq. 12 is also valid: made of N equally spaced segments, of length ∆x. Head, H,
and fluid velocity, V, are initially known for each of these
2
V + λg = V + a (12) sections from steady-state analysis. In the computation, Eqs.
λg 14 and 16 are used to determine the resulting time-step, and,
hence, the mesh size in the grid system. Along the character-
Solving for λ, we obtain λ = ± (a/g). Thus, these two real, istic curves, the time increment, ∆t, is related to the space in-
distinct values of λ convert the two partial differential equa- crement, ∆x, by: ∆t = ∆x/a.
tions into two total differential equations, subject to the re- By using a first-order approximation in the integration
strictions of Eqs. 8 and 9. along the C+ and C- characteristics, Eqs. 13 and 16 be-
Substituting for λ, we obtain Eqs. 13, 14, 15 and 16. Since come Eq. 17 along the C+ characteristic, and Eq. 18 along
in water-hammer calculations the value of V is small com- the C- characteristic:
pared with a, it may be dropped.
g f VA VA
g dH dV fV V a H P – H A + VP – V A + 2D ∆t = 0 (17)
a dt + dt + 2D = 0 (13)

dx = V + a ≅ a g f VB VB
dt
(14) a H P – H B + VP – V B + 2D ∆t = 0 (18)

g fV V
– a dH + dV + =0 (15) Adding Eqs. 17 and 18 eliminates HP, while subtracting
dt dt 2D eliminates VP, resulting in Eqs. 19 and 20, respectively:

dx = V – a ≅ –a
dt
(16) [
VP = 0.5 V A + V B + g/a H A – H B

The significance of these equations may be interpreted by


– f ∆t/2D V A V A + V B V B ] (19)
some considerations in the x-t plane (Figure 1). In this plane,
the curve labeled as C+ is a plot of Eq. 14, while C- is a plot [
H P = 0.5 H A + H B + a/g V A – V B
of Eq. 16. Equation 13 is valid only along the C+ characteris-
tic, while Eq. 15 is valid only along the C- characteristic.
Equations 13 and 15 contain two unknowns for any point on
– af ∆t/2gD V A V A – V B V B ] (20)

the characteristic, but at the intersection of the two curves C+

CEP November 2001 www.cepmagazine.org 45


Fluids/Solids Handling

25

4.0 s
20
Separator
3.0 s

Pressure, barg
0.6 s
15
1.0 s 2.0 s

Gas Liquid 10
1.4 s

5 0.2 s
1 2 3 4 5 6 7 8 9 10
Distance, m
0
0 14 28 42 56 70 84 98
■ Figure 3. Interface model for separator and downstream piping.
Pipe Length, m
From an analysis of these equations, it can be seen that the ■ Figure 4. Pressure profiles along the pipe length.
terms of the right-hand sides are constants and known values
from the previous time-step. The solution is then carried out 25
at the intersections of the characteristic curves, as shown in
Figure 1. The solution can be carried out only in a limited re- 20 4.0 s
gion, unless information is given for external conditions as a
Flluid Velocity, m/s 3.0 s
function of time for x = 0 and for x = L (at the initial and ter-
minal sections of the pipeline). Once the piezometric head 15
0.2 s
(pressure expressed as height of liquid column) and the fluid 1.4 s
velocity values are known for a given time-step, the proce- 10
dure is repeated for the next step, until the required time peri- 1.0 s
0.6 s
od is covered.
5
In standard water-hammer analysis two simplifying as-
sumptions are often made: 2.0 s
0
1. The bulk properties of the fluid are considered constant
0 14 28 42 56 70 84 98
throughout the pipe length.
2. The time-step is taken as the time interval required by a Pipe Length, m
sound wave to travel the length of the space step (∆t = ∆x/a).
This avoids the need for interpolation, since the space-time ■ Figure 5. Velocity profiles along the pipe length.
grid points are always on the characteristic curves.
In reality in our case, the pressure surge will cause
changes in the density and the bulk modulus of the dis- 4.5
persed gas phase, thus causing a variation in the speed of
propagation of the pressure wave. It is therefore necessary 4
to introduce local fluid properties (density and bulk modu-
lus) to evaluate the wave speed, and to select a fixed inte- 3.5
gration time-step not linked to the wave propagation speed
(which varies as the surge propagates along the line). Inter- 3
polation of the fluid velocity and head values have been in-
2.5
Times, s

troduced in the model, with the interpolation scheme illus-


trated in Figure 2.
2
The head in Points R and S is interpolated from the values
calculated in A, C, B using Eqs. 21 and 22: 1.5

H(R) = H(C) + H(A) – H(C) a∆t (21) 1


∆x
0.5
H(S) = H(C) + H(B) – H(C) a∆t (22)
∆x 0
0 12 24 36 48 60
The sonic velocity, a, is calculated based on the pressure Pipe Length, m
profile determined at the previous step. A similar procedure is
followed for the interpolation of the fluid velocity profile. ■ Figure 6. Position of the interface during the event.

46 www.cepmagazine.org November 2001 CEP


Boundary conditions normal operating conditions. This implies that the valve
Boundary conditions here refers to those conditions at the maintains a constant flow coefficient, CV, corresponding to
end section of a pipe segment. At either end of a pipeline, the one required to attain the design liquid flowrate at the
only one of the two equations, C+ or C-, is available, in the normal (steady-state) ∆P.
two unknowns, HP and VP (head and velocity). For a left- Equations 23 to 26 (as reported in the sizing manual of a
hand boundary, the C- curve holds, and for a right-hand major control valve supplier (3)) are used to characterize the
boundary, the C+ curve is valid. An auxiliary equation is flow through the control valve. Since the pressure piping ge-
needed for each case that specifies HP and VP, or some rela- ometry factor, FP, appears in both equations, it can be as-
tionship between the two, so that the equations can be solved sumed to be equal to unity without affecting the accuracy of
for the two unknowns. the calculation. The critical pressure ratio may be estimated
for a real gas by Eq. 27:
Interface model
The interface model establishes the boundary conditions W
C V, liq = (23)
at the interface where the gas is injected in the line owing to 27.3FP P1 – P2 γ
the loss of liquid level in the upstream separator, and the liq-
uid mass flowing along the pipe (Figure 3). The model evalu-
ates the pressure at the gas/liquid interface, while considering C V, gas = W
that the gas phase behaves as a perfect gas. If desired, correc- T 1Z (24)
94.8FPPV Y
tions for gas compressibility can be introduced. The pressure X PM
is supposed to be uniform in the whole gas volume down-
stream of the control valve. Its variation over time is evaluat- Y = 1 –[x/(3FkXT)] (25)
ed by calculating the total gas moles that have entered the
pipeline from the loss of the liquid level, and the total volume Fk = k/1.40 (26)
available for the gas due to the displacement of the interface
up to the time-step being considered. k

The calculation starts from steady-state conditions, so the Pc/P1 = 2 k–1


= XT
k+1 (27)
initial velocity of the interface is the bulk velocity of the liq-
uid in normal operation. The model is the same as would
apply for an upstream reservoir with pressure variation, only The flow will remain critical and will only depend upon
the boundary conditions apply to a mobile section, the inter- the upstream pressure, as long as the ratio between the
face. The location of the interface is calculated at each time- downstream and the upstream pressures stays lower than the
step, and the boundary conditions are applied at the pipe’s above value. The gas entering the pipeline at each time-step
longitudinal section that corresponds to the distance traveled is calculated by computing the gas flowrate through the
by the interface from the start of the integration up to the cur- valve with the steady-state valve CV, the normal operating
rent time-step. The boundary conditions are applied to the upstream pressure (separator pressure), and the downstream
subsequent section only after the interface has moved for the pressure calculated for the gas space at the previous step.
full ∆x length. The initial pressure is assumed to be the steady-state operat-
Pressure (head) and flowrate (velocity) are calculated for ing pressure downstream of the control valve in liquid ser-
the liquid along the pipeline. The pressure is assumed to be vice operation. The operating valve coefficient, CV, is com-
equal to the gas pressure at interface, and is the same for all puted on the basis of the selected steady-state liquid
sections of the pipeline that lay upstream of the current posi- flowrate, and the valve is considered to maintain the same
tion of the interface. As the position of the interface moves, coefficient throughout operation. This coefficient is used at
the volume available for the gas increases. The gas pressure each step to compute the gas flowrate.
is calculated based on the total gas moles in the pipeline up to The volume of pipeline made available for the gas during
the current time-step, and on the current volume. the time step ∆t is calculated as the product of the liquid
flowrate at the interface (m3/s) times the duration of the time-
Gas rate step(s). This volume is added to the total volume displaced
The rate of gas entering the pipeline is calculated based on by the gas in the previous steps, and the pressure is calculated
the normal operating pressure in the separator, the normal gas by applying the ideal gas law to the total gas moles occupy-
characteristics (molecular weight, compressibility factor, spe- ing the volume. At each integration step, the pressure down-
cific heat ratio), and the operating pressure downstream of stream of the valve is calculated by converging on the gas
the control valve. This pressure is initially assumed to be the flowrate via the secant method. The pressure values (poles)
steady-state value; subsequently, the pressure is calculated at assumed to initialize the computation, Pa and Pb, are selected
each time-step. It is assumed that the loss of liquid level in as the extreme values of the possible range: Pa = Psep (pres-
the separator occurs with the control valve blocked in under sure downstream of the valve equal to the upstream separator

CEP November 2001 www.cepmagazine.org 47


Fluids/Solids Handling

Table 1. Data for methanol example. Table 2. Influence of parameters on maximum peak pressure.
Parameter Value Maximum Pressure Developed in Letdown Circuit, barg
High-pressure separator pressure, barg 80 % Gas volume downstream control valve
Low-pressure flash vessel pressure, barg 5 0.5% 3% 15%
Liquid density, kg/m3 800
Initial liquid 20 m3/h 18.15 13.94 10.43
Liquid viscosity, cP 0.6
flowrate
Liquid bulk modulus, kg/m2 2 E+07 60 m3/h 28.75 21.76 16.69
Volume fraction of gas downstream control valve 0.005–0.15 100 m3/h 35.90 27.56 20.82
Liquid flowrate before loss of level, m3/h 20–100
Gas molecular weight 4
ly upon loss of level, and propagates with a step profile at
Gas specific-heat ratio 1.4
sonic speed along the pipe. When the whole pipe is pressur-
ized, a gradual variation, caused by friction losses, is estab-
Pipe characteristics lished between the interface front and the pipe outlet. The
Dia., mm 154 profile becomes steeper as the fluid is accelerated by the
pushing action of the gas.
Thickness, mm 7
• Only for large, initial gas-volume-fractions does the
Material Carbon steel pressure slightly increase during propagation. This can be ex-
Roughness, mm 0.05 plained as due to a slowdown of the interface displacement
Modulus of elasticity, kg/m2 2 E+10 speed due to the increase of fluid density caused by compres-
sion of the gas fraction.
Total length downstream valve, m 100
• The flow velocity of the fluid, after the propagation of
the pressure wave from the interface to the pipe outlet, in-
pressure, which implies zero flow across the valve) and Pb = creases gradually, with a bulk motion of the fluid in the pipe,
0 (pressure downstream of the valve equal to zero, which im- and reaches its maximum value when the interface reaches
plies maximum flow across the valve). At each step, the pre- the pipe outlet. The total displacement of the fluid occurs in a
viously calculated value of the pressure is taken as a new se- few seconds as the fluid velocity reaches high values.
cant pole, substituting the previous pole (P = Psep in the ini- • Maximum caution in design appears to be required in
tial step), until the values calculated in two subsequent cycles cases such as high-pressure gas solvent treating, due to the
are within a given tolerance (fixed at a relative error of 10–4 ). high density of the liquid and the relatively low amount of gas
being released by the solvent in the letdown process. These
Example and insights gained factors tend to involve higher pressures and larger thrusts on
A typical example is presented that evaluates the pressure supports, due to larger changes in density at the interface.
profiles in the letdown line for a raw condensate letdown cir-
cuit from a methanol synthesis loop. Typical conditions se- Computer program
lected are reported in Table 1. Figures 4 and 5 show the pres- A listing of a BASIC computer program is available on
sure and velocity profiles in the letdown piping circuits, eval- the author’s web page at http://web.tiscali.it/Ancihome. CEP
uated at different times after the gas breakthrough. Figure 6
traces the position of the interface as a function of time, dur-
ing the displacement of the initial fluid by the injected gas. Literature Cited
Parameters for these three figures are 3% flash and 60 m3/h.
1. Wylie E. B., et al., “Fluid Transients in Systems,” Prentice Hall, En-
The study was performed for various different gas vol- glewood Cliffs, NJ (1993)
ume-fractions in the liquid downstream of the control valve 2. Swaffield & Boldy, website: www.iteract.cam.ac.uk/wh/SWAFFIELD
(as calculated at steady-state conditions, before gas break- 3. “Control Valve Sizing and Selection Handbook,” Masoneilan Co., Bul-
through and surge occurrence), and different initial flow con- letin OZ 1000 (June 1994) (now Dresser Flow Control, Avon, MA).
ditions. The influence of these parameters on the peak value
of the pressure is summarized in Table 2. The following gen-
eral conclusions can be drawn:
GIANNI ANCI is a founder and managing director of EPSI, a process
• The calculated pressure and velocity profiles at various consulting company in Rome, Italy (E-mail: ganci@technip.com). He has
time intervals confirm the intuitive expectations: the presence nearly 30 years of experience in process synthesis, design and
of a larger gas fraction dampens the pressure surge, and re- engineering. He is the author of Italian Patent No. 1276517, “Simultaneous
duces the speed of propagation of the pressure front. Also, Production of Aviation Jet Fuel and Low Sulphur Gasoil by Hydrotreatment.”
His current commitments are with Technip Italy as process and engineering
the maximum fluid velocity is reduced with respect to lower manager in project management support contracts in Argentina for LNG
gas fractions. recovery at Compañia MEGA and in a methanol plant for Repsol-YPF. He is a
• The maximum pressure generally is reached immediate- graduate of Rome Univ., Italy, and a chartered professional engineer.

48 www.cepmagazine.org November 2001 CEP

You might also like