You are on page 1of 12

The hot-vapor bypass pressure control of distillation column: a

design review

Herbert Campos Gonçalves Teixeira

Process control engineer

Projetos de Refino, Gás e Eneregia (PRGE)

Petrobras S.A.

herbertteixeira@petrobras.com.br
The hot-vapor bypass pressure control of distillation column: a

design review

Abstract

The hot-vapor bypass has been designed for pressure control of distillation column for more than forty

years. The pressure balance between the top of the column and the overhead drum and the energy

balance into the overhead drum have been used to size the hot bypass valve. Nevertheless, the

overhead drum dimension has not been taken into consideration, even with the assumption that the hot

bypass vapor is totally condensed inside the overhead drum. The liquid surface area and the liquid

temperature inside the overhead drum determine the hot bypass vapor flow rate that is condensed inside

the drum and rule the energy balance into the overhead drum and the liquid temperature inside the

overhead drum. This paper describes the design criteria for sizing the hot bypass control system taking

into account the overhead drum size. Some problems are reported when these criteria have not been

considered.

Keywords
Pressure control

Distillation column

Hot-vapor bypass system


INTRODUCTION

An optimum distillation column design minimizes total cost: the fixed capital cost of the equipment

plus the variable cost of operation. Capital costs are the total fixed project costs to design, construct, and

start-up the process. Variable costs are the operating costs of utilities, raw material, and workforce.

There are several shortcut methods for distillation column designing. Conventional method can

design the distillation column taking into account the degrees of freedom available for design. The

degrees of freedom are the total number of variables less the total number of equations describing the

problem. The design of the distillation column has five degrees of freedom. There are five more variables

than equations. Usually, the procedure to design a distillation column first determines which variables will

be set and what variables will be calculated by the design. The method must therefore set five variables

to design a distillation column:

• The column pressure.

• The distillate and bottom products compositions.

• The column reflux ratio.

• The feed stage location.

Typical production objectives require delivery of products that have compositions within certain

specifications. Column pressure will be set by utility considerations. The design procedure also fixes a

minimum approach temperature in the reboiler and condenser. Given these conditions, an optimization

step will be follow the design to find what reflux ratio and feed stage location produce the most

economical design of the distillation column system. As a result, the designer finds the number of trays,

and condenser and reboiler duties for the column.

To the process control engineer a two product single distillation column typically has also five

degrees of freedom:

• The distillate and bottom product flow.

• The reflux flow.

• The heat input flow.

• The column pressure.

The process control objective is to regulate the product compositions in this process on a desired

set point. The control engineer first step to determine the pairing of control and manipulative variables of

distillation column is to reduce the number of degrees of freedom.


Two material inventories that must be controlled in the conventional distillation column are the

material in the overhead drum and column bottom. Therefore two degrees of freedom must be used for

these two control variables.

Ideally we should measure and control compositions to meet these objectives. Compositions are

measured by chromatographs. However, chromatographs tend to be expensive and in some applications

they can be difficult to maintain. For these reasons, composition analyzers sometimes are not preferred.

Instead an inferential variable, temperature, is widely used in distillation control applications.

In general, the composition, temperature, and pressure of a boiling mixture are thermodynamically

related. Column pressure has a direct influence on the separation capability of the column. The pressure

control significantly affects the variable cost, and it is necessary to take into consideration the

controllability and stability of process during column design, because they have an effect on the

equipment sizes and consequently on the capital cost. If the pressure is difficult to control, it will require

additional labor to operate and create higher utility and raw material costs due to upsets. Therefore, it is

important to hold the column pressure constant when the column are being designed and operated.

THE DISTILLATION COLUMN PRESSURE CONTROL METHODS

Chin1 observed that column pressure is one of the most difficult distillation variables to control. In his

paper, Chin presented a description of 21 different pressure control methods showing advantages,

disadvantages and applications.

Among these pressure control methods, the hot-vapor bypass is one of the most used when the

overhead vapors from the column are totally condensed. The hot-vapor bypass has been designed for

pressure control of distillation column for more than forty years. Whistler 2 reported that in 1954 the hot-

vapor bypass was a recent design practice. Chin commented that this method has been controversial for

several reasons. First, the merits of locating the condenser at ground level versus higher are hard to

quantify, although for maintenance, it means that inspection, bundle pulling, cleaning are very much

easier; and there is an attractive savings in steelwork, platforms and trolleys. Second, it is not easy to

understand how the method works. Third, the design is empirical. And finally, some installations have

failed; yet others are operating successfully. Sloley 3 and Lieberman4 presented some case of study and

troubleshooting with some hot bypass installations that have failed.

A typical arrangement for hot-vapor bypass on distillation column is shown in the figure 1. This

condenser is mounted at the ground level below the overhead drum. The condenser is submerged.
Flooding the condenser shell with condensate manipulates the heat transfer rate, in spite of the fact that it

requires more area than an elevated condenser of equal condensing capacity. This additional area

required makes the hot-vapor bypass condenser more expensive.

PCOL PIC

MBYPASS

MLCOND

PDRUM
MVAPOR T-1

H
x=0
P-1-- HDRUM
x

HCOND
TLCOND grade
level

Figure 1 - typical arrangement for hot-vapor bypass system on distillation column

THE CONVENTIONAL DESIGN METHOD FOR SIZE THE HOT-VAPOR BYPASS PRESSURE
SYSTEM

Durand4 described a method that determines the proper control valve size by estimating the amount

of vapors that must be bypassed in order to maintain the required pressure in the column. Basically, it

makes the energy balance in the overhead drum taking into account that the condensate is always

subcooled at the condenser outlet. Durand made the assumptions that the amount of vapor bypassed

must be enough to heat the subcooled condensate to the saturation temperature at overhead drum

pressure. Equation (1) reproduces the energy balance considering this assumptions:

M LCOND  c PL  (TDRUM − TLCOND )


M BYPASS = (1)
λ VAPOR

The pressure balance arrives that the pressure drop across the bypass valve is principally the

difference in static head between the overhead drum and the condenser levels, assuming the condenser

is near the grade. This results in the equation (2):


PCOL = PDRUM −  L  g  H − PCOND (2)

Durand remarked that the distance between overhead drum liquid level and grade usually is defined

by NPSH (net positive suction head) requirements of distillate pumps. Moreover, although the sub-cooling

depends on heat transfer conditions at condenser, the sub-cooling temperature is assumed to be equal 5

oC above inlet water temperature at the condenser. These assumptions, which are the empirical design

rules mentioned by Chin, will yield a conservative system design, over-sizing the bypass valve.

A REVIEW OF THE DESIGN METHOD: THE OVERHEAD DRUM ROLE

One of Durand premises is that the overhead vapors bypassed are totally condensed into the

overhead drum. Thus, the vapors bypassed enter the drum above the liquid level. The liquid surface in

the drum is cooler than the vapor saturation temperature. The method described by Durand does not take

into consideration the physical mechanisms of heat transfer and condensation that occur at the vapor-

liquid interface.

Condensation occurs when the temperature of a vapor is reduced below the dew point. The latent

energy of the vapor is released; heat is transferred to the liquid surface. There may be several

complicating features associated with this condensation. The liquid condensed at the top of the liquid

surface flows downward under the influence of gravity. A thin layer of condensed liquid is formed at liquid-

vapor interface, and the temperature, which is at this interface, is the saturation temperature TSAT.

However, the liquid under this layer is maintained at a temperature below the saturation temperature.

Despite the difficulties of the condensation at overhead drum, Tundidor8 showed that it guards

similarities with turbulent film condensation on a vertical plate described by Bayazitoglu 6. Tundidor noted

that the similarities come from the fact that the condensation takes place continuously over the vertical

plate which is kept cooled, the plate is covered with a thin layer of liquid, and the condensed liquid is

removed from the plate surface by the force of gravity. This situation is known as filmwise condensation.

Incropera7 wrote that Nusselt was the first that had analyzed the laminar filmwise condensation over a

vertical plate at a constant temperature. Bayazitoglu also underlined that Kirkbride had proposed the

empirical correlation valid for Re>1800 for film condensation on a vertical plate after start of turbulence,

expressed in equation (3).

1/ 3
  L2 
h m   2 
 = 0.0077  (Re)0.4 (3)
 kL  L  g 
2
Departing from equation (3) Tundidor showed that the total mass flow rate of condensate at liquid-

vapor interface could be obtained by the following equations:

Q = A L  h m  (TSAT − TDRUM ) (4) And

Q = M VAPOR   VAPOR (5)

From equations (3), (4) and (5), Tudidor arrived in equation (6).

   g   L2 A L  (TSAT − TDRUM )
1/ 3
kL 
M 0,6
= 0,0134      (6)
VAPOR  (  L )0 ,4    2   VAPOR
 L   L 

Where:

AL = L  D (7)

The equation (6) indicates that the mass flow rate of condensate at liquid-vapor interface depends

on the area of the interface, the sub-cooling temperature of the liquid inside the overhead drum and

physical properties of the condensate. One of the assumptions adopted by Durand was that the

temperature of the liquid inside the overhead drum would be at the saturation temperature. However, the

equation (6) would result that there would be no mass flow rate of condensate at liquid-vapor interface,

since T is equal zero. This contradiction leads us that the temperature of the liquid inside the overhead

drum must be subcooled. When the temperature of drum is at saturation temperature; the bypass valve

should be fully closed to maintain the pressure constant.

In other words, Tudidor showed that the overhead drum has one important role in hot-vapor bypass

system design, because the size and the temperature liquid inside the overhead drum determines the

amount of vapor that must be bypassed.

A REVIEW OF THE DESIGN METHOD: THE DYNAMIC POINT OF VIEW

So far, The equations (1) to (6), which are utilized for hot-vapor bypass system design, were

obtained by using the formulas commonly used for steady pipe flows. But, its dynamic response is

sometimes of interest. In fact, we are interested in how the pressures of column and overhead drum vary

with time.

In the hot-vapor bypass configuration, the column pressure PCOL and drum pressure PDRUM are

exerted by a gas whose inertia and viscosity may be considered negligible compared with those of liquid

into pipe and condenser. If the pressures vary with time, the flow of liquid and the pressure drop due to

friction across the pipe and condenser vary with time. The variation of liquid flow rate in the pipe and
condenser is caused by action of various forces. Equation (8) represents the force balance considering

the liquid in the pipe and condenser as free body and taking into account the following forces:

1. The gravity force (weight) distributed uniformly over the whole body of fluid.

2. The force due to the wall shearing stress related to the motion of the liquid.

3. The forces on the two ends of the free body due to pressures PCOL and PDRUM.

4. The effective mass of moving liquid. Based on the kinetic energy we can arrive that the effective

mass of moving liquid is four-thirds of the actual mass in the pipe and condenser.

d2 d2 1
   P −      g  x −   d  Z   0 =    d 2  Z    x (8)
4 2 3
Where:

P = PCOL − PDRUM (9)

Z = H COND − H DRUM (10)

The force per unit area is known as the shear stress (). The turbulent wall shearing stress is

given by equation (11):

0,0535   L0,75   L0, 25  x 1,75


τO = (11)
d 0,25
The most significant feature of equation (11) is that it represents a nonlinear relation between

shear stress and liquid velocity in pipe and condenser. When the force due to wall shearing stress is

substituted in equation (8) results is a nonlinear differential equation because of the term x
1, 75
. This

nonlinear equation cannot be solved analytically. Nevertheless, Doebelin 8 reported that engineers

have developed a useful and practical method of analysis, which have been verified in experimental

tests. This approach is based on the observation that the linearization of term x
1, 75
does not result

in a form of oscillation radically different from the nonlinear approach if the pressure disturbances

are small. The linearization reduces to equation (12):

2 0,107   0, 25  Z  VAV 0,75


P
 x +  x + x =
(3  g Z)
(12)
d  g 
1,25 0 , 25
2 g  

In Laplace form, this becomes the equation (13) and (14):


 0,107   0, 25  Z  VAV 0,75
 P(s)
  X(s) =
2
  s 2
+  s + 1
 (3  g Z) d  g 
1,25 0 , 25  2 g  
(13)
 

X(s ) 1 (2  g   )
=
P(s )
(14)
2 0,107   0, 25  Z  VAV 0,75
 s2 + s +1
(3  g Z) d 1,25  g   0, 25

Frequent use of second-order equation is made in characterizing different dynamics of process

systems. The shape of the transient response can generally be characterized satisfactorily by two

parameters, a frequency and a damping coefficient. Equation (14) can be rewrite as a second-order

form:

X(s ) K
=
P(s )
(15)
1 2
 s2 + s +1
 n2 n
Thus, the gain, frequency and damping coefficient are:

1
K= (16)
2 g  

3 g
n =
2 Z (17)

0,0655   0, 25  VAV0,75
Z
 =  (18)
d 
1,25 0 , 25
g

The significance of n and  becomes apparent after considering a step change in pressures input

P. With a damping coefficient less than 1, the output (x), which means the submerged tube of

condenser, overshoots the final value and oscillates before coming to equilibrium. The system is known

as an underdamped system. The equations (17) and (18) show us that the hot-vapor bypass system can

easily be underdamped. To avoid oscillations, we can change two parameters in equations (17) and (18),

Z and VAV. Z is the distance between overhead drum and condenser liquid level, and as Durand remarked

Z is a requirement of distillate pumps.

VAV is the average velocity of the liquid in pipe and condenser shell. Therefore, it is related to

pressure drop across the pipe and condenser shell. In others words, if the pressure drop across the pipe
and condenser shell increase, which implies that VAV also increase, the damping coefficient will increase,

which means that the hot-vapor bypass system will have a dynamic response with less oscillations. As a

result, the pressure drop across the pipe and condenser shell is an important variable to sizing the hot-

vapor bypass system.

Furthermore, lets examine the hot-bypass system at minimum required capacity. Minimum capacity

means that more tubes of the condenser are submerged. So, Z increases. The pressure across the

condenser would decrease, and the temperature at condenser outlet and in the overhead drum would be

lower because the condenser has more tubes submerged. Thus when the condenser operates at

minimum capacity the hot-vapor bypass becomes more underdamped. Moreover, the equation (06)

indicates that the flow rate of vapor bypassed would increase because the temperature and in the

overhead drum decreased.

CONCLUSIONS

Many assumptions were made in the above analysis. The formulas for steady state were

employed for an unsteady situation. Turbulent equations were used for the analysis and the nonlinear

differential equation actually has no analytical solution. Thus what is the meaning of n and  attached to

the hot-vapor bypass process? A more complete theory and some supporting data shall be made.

Despite the fact of the simplified analysis, some important conclusion can be taken from the

above appraisal:

• The size and subcooled temperature of liquid of the overhead drum determines the mass flow rate

that shall be bypassed.

• The liquids inside the overhead drum and at condenser outlet are not in equilibrium with vapor, but

they are subcooled. The temperature at condenser outlet is lower than temperature in the overhead

drum.

• The pressure drop across the pipe and condenser shell is an important parameter. It shall be as

bigger as possible to avoid oscillating response.

• It is important to examine the hot-bypass system at minimum required capacity, as well as in nominal

capacity, since at minimum capacity the hot-vapor bypass becomes more underdamped than at

nominal capacity.

NOMENCLATURE

AL liquid surface area at overhead drum, m2 .


cPL Condensate heat capacity, kcal/kg oC.

d Pipe equivalent diameter, m.

D Overhead drum diameter, m.

g acceleration due to gravity, m/s2 .

hm Heat transfer coefficient, W/(m.oC).

COND Condenser level height from grade level, m.

DRUM Overhead drum level height from grade level, m.

 Distance between overhead drum and condenser levels, m.

kL Thermal conductivity of liquid, W/(m.oC).

K Process gain, m2/(kg.s).

L Overhead drum length, m.

MBYPASS Mass flow rate of bypassed vapor, kg/h.

MLCOND Mass flow rate of condensed vapors at condenser outlet, kg/h.

MVAPOR Mass flow rate of condensed vapors at vapor-liquid interface, kg/h.

PDRUM Overhead drum pressure, kgf/cm2 g.

PCOL Overhead column pressure, kgf/cm2 g.

PCOND Pressure drop across the condenser, kgf/cm2 g.

Q Heat transfer rate, kcal/h.

Re Reynolds number.

s Laplace variable.

TDRUM Overhead drum liquid temperature, oC.

TLCOND Subcooled condensate temperature at condenser outlet, oC.

TSAT Saturation temperature of vapor, oC.

VAV Average velocity of liquid in pipe and condenser, m/s.

x distance of condenser level from inlet nozzle, m.

Z Pipe length with liquid (including the condenser level), m.

VAPOR Overhead drum vapors latent heat, kcal/kg.

L Overhead drum liquid density, kg/m3.

L Viscosity of liquid, kg/(m.s).


0 Wall shearing stress, N/m2.

n Frequency, rad/s.

 Damping coefficient.

REFERENCES

1. WHISTLER, Arthur M. Locate condenser at ground level. Petroleum Refiner, Vol. 33, No. 3, March

1954, p. 173-174.

2. CHIN, T. G. Guide to distillation pressure control methods. Hydrocarbon Processing, October 1979,

p. 145-153.

3. SLOLEY, Andrew W. Troubleshooting with exchanger liquid levels. AIChE Spring Meeting:

Symposium on Industrial Applications in Process Heat Transfer, New Orleans - Louisiana – 8-12

March 1998.

4. LIEBERMAN, Norman P and LIEBERMAN, Elizabeth T. A working guides to process equipment: How

process equipment works. 1 ed. McGraw-Hill, Inc: New York, 1997. Cap. 13. p. 147-161.

5. Durand, Alejandro A. Sizing hot-vapors bypass valve. Chemical Engineering, August 25, 1980, p.

111-112.

6. BAYAZITOGLU, Yildiz and Özisik, M. N. Elements of heat transfer. International edition. McGraw-Hill,

Inc. New York, 1988. Cap. 10. p. 303-326.

7. INCROPERA, Frank P. and DeWITT, David P. Fundamental of heat and mass transfer. 3a ed. John

Wiley & Son, Inc. Singapore, 1990. Cap. 10. p. 587-637.

8. TUNDIDOR, Miguel A. Modelo para cálculo do “hot bypass”. Unpublished report. Revisão 0.

Petrobrás/CENPES. 07/02/1996.

9. HOLLAND, F. A. Fluid flow for chemical engineers. 2a ed. Edward Arnold Publishers Ltd. London,

1980.

You might also like