Distillation of the “Light Ends” from Crude Oil in Petroleum Processing

Abstract

The “light ends” unit is the only process in a refinery configuration that is designed to separate “almost” pure components from the crude oil. Its growth initially resulted from the need of those components such as butanes and propanes to satisfy a market of portable cooking fuel and industrial fuels. The need for modern gasolines has added to the demand for light ends to make high-octane blending components. This chapter describes the process and design of units to separate the light ends into useful products or into cuts for further processing in the refinery. An example design is provided.

David S. J. Jones: deceased.

Introduction

The “light ends” unit is the only process in a refinery configuration that is designed to separate “almost” pure components from the crude oil. Its particular growth has resulted from the need for those components such as butanes and propanes to satisfy a market of portable cooking fuel and industrial fuels. That these products can be suitably compressed and stored in small, easily handled containers at ambient temperatures provided the market popularity for these products, suitably titled butane LPG and propane LPG. The term LPG refers to liquefied petroleum gas.

The introduction of “no lead” gasoline in the late 1960s established the need of octane sources additional to the aromatics provided by high severity catalytic reforming. A source of such high-octane additives is found in some isomers of butane and pentane. This added to the need for light ends processes which in many cases included the separation of isobutanes from the butane stream and also isopentanes from the light naphtha stream.

A Process Description of a “Light End” Unit

The “light ends” of crude oil are considered to be those fractions in the crude that have a boiling point below cyclohexane. The “light ends” distillation units however include the light naphtha cut separation, which is predominately pentanes and cyclopentanes, from heavy naphtha which contains the hexanes and heavier hydrocarbons necessary for the catalytic reformer feed. The feed to the “light ends” distillation process is usually the full range naphtha distillate from the atmospheric crude distillation unit overhead condensate drum. In many cases the distillates from stabilizing cracker and reformer products are added to the crude unit overhead distillate to be included in the “light ends” unit feed.

A typical process configuration for this unit is given in the flow diagram (Fig. 1).

Fig. 1

A typical light ends unit configuration

In this configuration the total feed to the unit is debutanized in the first tower. The butanes and lighter hydrocarbons are totally condensed and collected in the column’s overhead drum. Part of this condensate is returned to the tower top tray as reflux. The remainder is routed to a depropanizer column. The bottom product from the debutanizer is the full range naphtha product. This enters a naphtha splitter column where it is fractionated to give an overhead distillate of light naphtha and a bottom product of heavy naphtha.

The depropanizer separates the debutanizer overhead distillate to give a propane fraction as an overhead distillate stream and the butane fraction (butane LPG) as the bottom product. The overhead distillate is fractionated in a deethanizer column to produce a rich propane stream (propane LPG) as the bottom product. The overhead from this column is predominately hydrocarbons lighter than propane. This stream is only partially condensed to provide reflux for the tower. The uncondensed vapor is normally routed to the refinery’s fuel gas system.

The products from the “light ends” unit are as follows:

• Naphtha splitter – light naphtha (overhead distillate)

• Heavy naphtha (bottom product-reformer feed)

• Depropanizer – butane LPG (bottom product)

• Deethanizer – propane LPG (bottom product)

• Fuel gas (overhead vapor)

Developing the Material Balances for Light Ends Units

In light ends towers, the material balance is developed on a molal balance. This type of balance is determined by the degree of separation of the feed molal components that enter the distillate fraction and those that leave with the bottom product.

Effective separation by fractionation in light ends towers obeys the same laws as those in the crude distillation units, that is: the degree of separation is the result of the number of trays (or stages) and the reflux (or overflow) in the column.

In the crude unit, this separation was measured by the difference between the ASTM 95 % point of the lighter fraction and the 5 % ASTM point of the heavier fraction. This is the ASTM gap or overlap.

In light ends towers the degree of separation is a little more precise. This is determined by the distribution of key components in the two fractions to be separated. Key components may be real components (such as C₄s or C₅s) or pseudo-components defined by their mid-boiling points. Normally key components are adjacent components by boiling point in the feed composition. Any two key components may be selected – a light key and a heavy key. By definition, the light key has the lower boiling point. Both key components must, however, be present in the distillate and bottom products of the column. If a side stream exists, then these keys must also be present in the side stream product.

There are several correlations that describe the behavior of these key components in their distribution and relationship to one another. By far, the more common of these correlations is the Fenske equation which relates the distribution of key components at minimum trays with infinite reflux. The equation is relatively simple and does not require iterative calculation techniques to solve it. The Fenske equation is:

$$ {N}_{\mathrm{m}+1}=\frac{\mathrm{Log}\left[\left({\left(\mathrm{L}\mathrm{T} \mathrm{key}/\mathrm{H}\mathrm{Y} \mathrm{key}\right)}_D\times {\left(\mathrm{H}\mathrm{Y} \mathrm{key}/\mathrm{L}\mathrm{T} \mathrm{key}\right)}_W\right)\right]}{\mathrm{Log}\frac{K_{LT\ \mathrm{key}}}{K_{\mathrm{HY}\ \mathrm{key}}}} $$

(1)

where:

• N _m = minimum number of theoretical trays at total reflux. The +1 is the reboiler which is counted as a theoretical tray.

• LT key = is the mole fraction of the selected light key.

• HY key = is the mole fraction of the selected heavy key.

• D = fractions in the distillate product.

• W = fractions in the bottom product.

• K _{LT key} = the equilibrium constant of the light key at mean system condition of temperature and pressure.

• K _{HY key} = the equilibrium constant of the heavy key again at mean system conditions.

The ratio of the equilibrium constants is called the “relative volatility” of the keys. Setting fractionation requirements for light ends towers is usually done to meet a product specification. More often than not, this specification is the vapor pressure of the heavier fraction or the tolerable amount of a heavy key allowed in the lighter fraction.

Sometimes, however, a specification for the separation may not be given. Under this circumstance, some judgment must be made in determining the most reasonable separation that can be achieved with the equipment. This item of the handbook addresses calculation techniques that satisfy either premise.

It is common to use process simulation software (HYSYS, Pro/II, and others) to perform the design calculations for light ends units. This allows case studies to optimize the process design. The details of the computer simulations are beyond the scope of this handbook. The manual procedure for the calculations is described here. It is helpful to understand the manual approach to establish the background and limitations behind the computerized calculations.

Light Ends Specifications Cases

• Case 1: Setting separation requirements to meet a specification

  In this case it is required to determine the amount of butanes that can be retained by a light naphtha cut to meet a RVP specification. The steps are as follows:

  • Step 1. Calculate the properties of the C₅ + naphtha from a component breakdown. These properties should give weight and mole rates per hour.

  • Step 2. Carry out a bubble point calculation of the C₅ + fraction and inserting butane as the unknown quantity x.

  • Step 3. The equilibrium constant used (K) in the calculation will be at the temperature and pressure conditions of the RVP (i.e., normally at 100 °F which is the test temperature).

  • Step 4. Either the equilibrium constants given in the charts in Appendix 3 of the chapter entitled “Process equipment for petroleum processing” in this handbook or the relationship of vapor pressure divided by the total system pressure may be used.

  • Step 5. By definition, the total moles of liquid given is equal to the total moles of vapor (calculated) in equilibrium at the bubble point. Thus equate and solve for x as the quantity of butanes tolerable to meet RVP.

• Case 2: Setting fractionation where no specification is given

  • Step 1. Determine the composition of the feed in terms of real components, pseudo-components, or both. Calculate this in moles/h and mole fractions.

  • Step 2. Select the key components, and decide the minimum distribution of one or other key. For example, in a debutanizer C₅s allowed into LPG must not be more than 2 % of the total C₄s and lighter. This is to protect the Butane LPG “Weathering” Test Specification.

  • Step 3. Give values to the fraction of LT and HY keys in the distillate and bottoms. Use x as the unknown where appropriate.

  • Step 4. The Fenske equation will be used to calculate the distribution of keys. Determine the value of N _m by taking the actual number of trays and using an efficiency of 70–75 % to arrive at total theoretical trays. Divide this figure by 1.5 to arrive at the minimum theoretical trays N _m. Do not forget to add 1 for the reboiler to use in the equation.

  • Step 5. Estimate the mean tower conditions. This can be achieved by examining past plant logs and other sources. Determine K values for the keys at this mean tower condition. Use published data for real components and the ratio of vapor pressure divided by total systems pressure for pseudo-components.

  • Step 6. Solve for x in the Fenske correlation. This will be the split of the key components and the basis for the material balance.

Very often when setting up a design for a light ends tower, the actual number of trays is not known at the time when calculation number two is required. The rules of thumb in Table 1 may be used as a guide.

Table 1 Rules of thumb for number of light ends column trays

Example Case

An example for illustrating Case 1 above is given in the chapter entitled “Introduction to Crude Oil and Petroleum Processing” of this handbook where the amount of butane LPG allowed to meet gasoline RVP is calculated. An example of Case 2 is given as follows:

The overhead distillate from an atmospheric crude distillation unit operating at 50,000 BPSD of Murban crude has the composition noted in Table 2.

Table 2 Full range naphtha composition

The key components for the debutanizer will be nC₄ and iC₅ as LT and HY keys, respectively. In the Fenske equation let x be the moles/h of nC₄ in the distillate. To satisfy the weathering test for butane LPG, the maximum amount of C₅s allowed in the debutanizer distillate is 2.0 mol% of the total C₄s and lighter. A reasonable number of actual trays in a debutanizer is 30. Allowing an efficiency of 70 %, the number of theoretical trays will be 21. It is reasonable to predict that the minimum theoretical trays will be the actual theoretical number divided by 1.5. Then N _m will be 21/1.5 = 14 adding one for the reboiler giving N_m+1 = 15.

From past data, average operating conditions of temperature and pressure for a debutanizer are 210 °F and 110 psig. At these conditions, the equilibrium constants for both keys are read from curves given in the GPSA Engineering Data as:

$$ \begin{array}{c}\mathrm{n}{\mathrm{C}}_4=1.48\\ {}\mathrm{i}{\mathrm{C}}_5=0.94\end{array} $$

Then the relative volatility ϕ = 1.48/0.94 = 1.57.

Using the Fenske equation:

$$ \begin{array}{c}{N}_{\mathrm{m}+1}=\frac{\mathrm{Log} \left.\left({\left(\mathrm{L}\mathrm{T}\;\mathrm{key}/\mathrm{H}\mathrm{Y}\;\mathrm{key}\right)}_D\times {\left(\mathrm{H}\mathrm{Y}\;\mathrm{Key}/ \mathrm{L}\mathrm{T}\;\mathrm{key}\right)}_W\right)\right)}{\mathrm{Log}\phi}\\ {}15=\frac{\mathrm{Log}\left(\left(x/3.8\times 112.1/108-x\right)\right)}{\mathrm{Log} 1.57} \\ {}15\times 0.196=\mathrm{Log}\left((112.1x)/\left(410.4-3.8x\right)\right) \\ {}871=112.1x/\left(410.4-3.8x\right) \\ {}x=104.48\;\mathrm{moles}/\mathrm{h}. \end{array} $$

Associated with nC₄ in the bottom product will be an equilibrium amount of iC₄. Although small, this will have an effect on the bottom product bubble point and therefore the tower bottom temperature. The amount of this iC₄ in the bottom product can be calculated using a similar method as that for the split between nC₄ and iC₅; thus the keys for this calculation will be iC₄ and nC₄. Let x be the moles/h of iC₄ in the bottom product. Then:

	Feed	Dist	Bottoms
iC4	39	x	39 – x
nC4	108	104.48	3.52

Using the Fenske equation again:

$$ \begin{array}{c}15=\frac{ \log \left(\left(x/104.48\right)\times \left(3.52/39-x\right)\right)}{ \log \left(K\mathrm{i}{\mathrm{C}}_4/K\mathrm{n}{\mathrm{C}}_4\right)}\\ {}=\frac{ \log \left(\left(x/104.48\right)\times \left(3.52/39-x\right)\right)}{ \log 1.3}\\ {}1.709= \log \left(\left(x/104.48\right)\times \left(3.52/39-x\right)\right)\end{array} $$

Then:

x = 38.99 mol/h. This gives zero iC4 in the bottom cut.

The material balance for the debutanizer can now be written and is given in Table 3.

Table 3 Debutanizer material balance

The material balances over the naphtha splitter, depropanizer, and the deethanizer follow the same technique in identifying key components and utilizing the Fenske equation.

Calculating the Operating Conditions in Light Ends Towers

Light ends units follow two calculation procedures for setting the conditions of temperature and pressure in the fractionating towers. The first procedure relates to towers in which the overhead product and reflux are totally condensed. The second procedure relates to those towers where the overhead product is not totally condensed.

Calculating the Tower Top Pressure and Temperature for Totally Condensed Distillate Product

This procedure commences with setting a realistic reflux drum temperature. This is fixed by the cooling medium temperature, such as ambient air temperature (for air coolers) or cooling water. As total condensation is required, then the pressure of the reflux drum must be the bubble point of the distillate product (and reflux) at the selected drum temperature. Once the drum pressure has been calculated, the tower top pressure can be determined by taking into account estimated or manufacturer’s specified pressure drops for equipment and piping between the drum and the tower top. As a rough estimate, condensers and/or heat exchangers in the system have between 3 and 5 psi pressure drop. Allow also about a 2 psi pressure drop for piping.

The tower top temperature is calculated as the dew point of the distillate product at the total overhead pressure. There are usually no steam or inert gases present in the light ends tower overheads, so total pressure may be used. The following example uses the debutanizer overhead from the material balance given in section “Developing the Material Balance for Light Ends Units” above.

The ambient air temperature for the site is 60 °F, and the operating temperature for the reflux drum will be set at 100 °F. The bubble pressure at this temperature is calculated and given in Table 4.

Table 4 Debutanizer reflux drum pressure. Distillate bubble point calculation at 100 °F (Trial 1)

Allowing 3 psi pressure drop over the overhead condenser and 2 psi for the associated overhead piping, the tower top pressure becomes 130 psia (115 psig). The temperature of the tower top is the dew point of the distillate product at the top pressure. This is given in Table 5.

Table 5 Tower top temperature

There is reasonable agreement that total y = total x; therefore the dew point and the tower top temperature at 115 psig is 152 °F.

Calculating the Overhead Conditions for Partially Condensed Distillate and Product

There are two circumstances where the overhead stream from a light ends tower may not be totally condensed. The most common of these is in the case of the deethanizer. Here, usually, only sufficient overhead stream is condensed to provide the overhead reflux stream. The reason for this is that at a normal condensing temperature, the pressure required at the reflux drum would be unacceptably high. It would be so high that the tower bottom pressure required would be higher than the bottom product’s critical pressure, and thus fractionation would not be possible. In this case the tower pressure is set again by the reflux drum pressure. Unlike the case of total condensation, the reflux drum pressure for this partial condensation is found as the dew point at the condensing temperature of the distillate product. In fact the reflux drum becomes a theoretical fractionation stage. In certain cases refrigeration is used to totally condense the overhead stream at an acceptable pressure, but this selection would be as a result of a study of the operation’s economics.

In certain processes where a high tower pressure (and consequently high temperature) may cause deterioration of one or more of the products from the tower, the tower pressure is reduced by condensing only a fraction of the overhead distillate product. For this purpose, a vaporization curve for the distillate product is constructed at a selected reflux drum temperature. This curve is developed by calculating a series of equilibrium compositions of the distillate product at the reflux drum temperature but over a pressure range. A reflux drum pressure can then be selected from the curve that satisfies an acceptable tower pressure profile. A final equilibrium calculation is then made at this pressure to provide the component and quantity composition of the liquid and vapor streams leaving the reflux drum.

An example of deethanizer overhead operating conditions is given below:

Assume that the overhead vapor from the reflux drum, which is the product in this case, has the following molar composition:

Mole	Fraction
C1	0.063
C2	0.443
C3	0.494

A reasonable temperature for operating the reflux drum is 100 °F (based on local ambient conditions). The reflux drum pressure is calculated from the dew point of the vapor at the reflux drum pressure, thus:

	Vapor mole fract y	K at 100 °F and 350 psia	Liquid mole fract x
C1	0.063	7.0	0.009
C2	0.443	2.0	0.222
C3	0.494	0.64	0.772
Total	1.000		1.003

This was the second trial starting with the pressure at 450 psia. The total x value is close enough to y to allow the pressure to be 350 psia.

Now the tower top conditions will be at the dew point of the reflux plus the product leaving the tower. The pressure will be the reflux drum pressure plus, say, 7 psi for the condenser and piping pressure drop. In this case this pressure will be 357 psia. Set the reflux ratio (moles reflux/mol product) to be 2.0:1 where the moles/h rate of the product is 7.9. The composition of the reflux stream is the x value of the product vapor. The dew point calculation to establish the tower top temperature therefore will be as shown in Table 6.

Table 6 Calculating the tower top temperature for a deethanizer

Note this dew point calculation may change if subsequent calculations show that the reflux ratio required will be significantly different to the one assumed here.

It is not proposed to show the case of the partially condensed product here. This calculation would be similar to the deethanizer and in this case the reflux composition will be the liquid phase from the equilibrium flash of the product.

Calculating the Tower Bottom Conditions for Light Ends Towers

The calculation to establish the tower bottom conditions for all light ends towers is the same. Only the values for pressure and composition change. The temperature of the product leaving the bottom of the tower will be at its bubble point at the tower bottom pressure. Consider the debutanizer column whose material balance was given in Table 2.

The number of actual trays (estimated) to accomplish the separation between the C₄s and the C₅s is 30. The pressure drop for fully loaded trays can be taken as being between 0.15 and 0.2 psi. Assume then a pressure drop of 0.17 psi per tray. The tower top pressure calculated earlier is 125 psia, and then the bottom pressure is 125 + (0.17 × 30) = 130 psia (Table 7).

Table 7 Debutanizer bottom temperature (bubble point)

Calculating the Number of Trays in Light Ends Towers

For definitive design work, one of the many excellent simulation packages should be used. However most simulation packages require good quality input data. This often means a fairly accurate estimate of the number of theoretical trays and the liquid/vapor traffic in the tower. An acceptably accurate “short-cut” method to arrive at the number of theoretical trays is given here. The estimate of the liquid/vapor traffic is discussed later in this chapter.

The Short-Cut Method for Predicting Number of Theoretical Trays

Three calculations or relationships are used to determine the number of theoretical trays in this method. These are:

• The Fenske calculation to determine the minimum number of trays at total reflux

• The Underwood calculation to determine the minimum reflux at infinite number of trays

• The Gilliland correlation which uses the result of the two calculations to give the theoretical number of trays

The Fenske Equation

This has been discussed earlier under the section dealing with the “Material Balance for Light Ends Towers.” The equation is as follows:

$$ {N}_{\mathrm{m}+1}=\mathrm{Log}\left[{\left(\mathrm{L}\mathrm{T}\;\mathrm{key}/\mathrm{H}\mathrm{Y}\;\mathrm{key}\right)}_D{\left(\mathrm{H}\mathrm{Y}\;\mathrm{key}/\mathrm{L}\mathrm{T}\;\mathrm{key}\right)}_W\right]\div \mathrm{Log}\left({K}_{\mathrm{LT}\ \mathrm{key}}/{K}_{\mathrm{HY}\ \mathrm{key}}\right) $$

(2)

where:

N _m = minimum number of theoretical trays at total reflux. The +1 is the reboiler which is counted as a theoretical tray.

LT key = is the mole fraction of the selected light key.

HY key = is the mole fraction of the selected heavy key.

D = fractions in the distillate product.

W = fractions in the bottom product.

K _{LT key} = the equilibrium constant of the light key at mean system conditions of temperature and pressure.

K_{HY key} = the equilibrium constant of the heavy key again at mean system conditions.

The Underwood Equation and Calculation

The Underwood equation is more complex than the Fenske and requires a trial and error calculation to solve it. The equation itself is in two parts: the first looks at the vapor volatilities (ratio of the Ks) of each component to one of the keys and then by trial and error arriving at an expression for a factor B that forces the equation to zero. This first equation is written as follows:

$$ {\displaystyle \sum \left.1\left(\left(\phi i\right)\cdot (xiF)\div (xiF)-B\right)\right)=0} $$

(3)

where:

• ϕi = the relative volatility of component i

• xiF = the mole fraction of component i in the feed

• B = the factor that forces the expression to zero

The second part of the equation is expressed as follows:

$$ {R}_{\left(\mathrm{m}+1\right)}={\displaystyle \sum \left(\left(\phi i\right)\cdot (xiD)\right)\div \left.1\left((xiD)-B\right)\right)} $$

(4)

where:

• R _m = minimum reflux at infinite number of trays

• xiD = the mole fraction of i in the distillate

The relationship between the Fenske equation and the Underwood is given by the Gilliland correlation shown in Fig. 2.

Fig. 2

The Gilliland correlation for calculating theoretical trays

An Example of the Underwood Equation Calculation

Consider the material balance of a debutanizer developed in Table 2.

Part 1 of the Underwood equation is calculated as shown in Table 8.

Table 8 Underwood equation, Part 1, calculation example

Part 2 of the Underwood equation is calculated as shown in Table 9.

Table 9 Underwood equation, Part 2, calculation example

Using the Gilliland Curve: $\frac{\left(R-{R}_{\mathrm{m}}\right)}{R+1}=0.27$

And from the curve: $\left(\frac{N-{N}_{\mathrm{m}}}{N+1}\right)=0.4.$

Then number of theoretical trays N will be N – N _m = 0.4 N + 0.4.

N _m calculated from the Fenske equation (see section on “Developing the Material Balance for Light Ends Units”) is 14.

Then N = 24.

Assume an average tray efficiency of 70 % then total actual trays = 34.

Condenser and Reboiler Duties

Both the condenser and reboiler duties are the result of the light ends tower heat balance that meets the degree of separation required in the products. It may be said that the condenser duty and operation determine the amount of wash liquid flow in the tower to meet the degree of rectification required for the lighter product. The reboiler in turn generates the vapor flow in the tower to satisfy the degree of stripping required for the particular separation and the quality of the heavy product. At a constant feed temperature and pressure, changes to either the condenser duty or the reboiler duty will affect the duty of the other. In other words the tower must always be in heat balance.

Calculating the Condenser Duty

The duty of the overhead condenser is determined by a heat balance over the tower top (above the top tray) and reflux drum, with the condenser duty being the unknown quantity. Using the data already determined for the debutanizer in the previous sections, the following is a calculation to determine the condenser duty for this unit.

Consider the following heat balance sketch (Fig. 3).

Fig. 3

The overhead heat balance figure

In the following heat balance, the liquid overflow from the top tray is based on the internal reflux ratio obtained from the Fenske, Underwood equations, and the Gilliland correlation. This ratio is in moles and its molecular weight is derived from the molal composition of the dew point calculation to establish the tower top temperature. It is the mole weight of the liquid in equilibrium with the overhead distillate vapor. Thus the total moles of the overflow liquid are 4.3 × 191.66 = 824.14 mol/h. Its mole weight is 56.79. Therefore overflow liquid is 46,803 lb/h. The heat balance now follows (Table 10).

Table 10 Tower top heat balance example

The top tray temperature is taken as 3° F above the tower top temperature. The vapor to the top tray is taken to be about 5° F above the top tray.

Condenser duty is 7,020,000 Btu/h.

Calculating the Reboiler Duty

It is important that the reboiler duty achieve a balance between generating an effective rate of vapor for stripping the bottom product while maintaining an economic vapor load on the stripping trays. As a rule of thumb, this can be achieved with a stripping rate as a percent of bottom product of between 70 and 90 mol %. The tower bottom temperature has already been calculated as 358 °F at 138 psia. The composition of the vapor in equilibrium at these conditions has also been calculated. Thus the moles/h of bottom product are 1,359.79; then the moles of strip-out required will be 1,359.79 × 0.9 = 1,223.8 mol/h. The molecular weight of the strip-out is 88 (from the equilibrium composition). The heat balance to determine the reboiler duty now follows. In this calculation the draw-off tray to the reboiler is about 10° F lower than the tower bottom temperature, that is, 348 °F (Table 11).

Table 11 Bottom tray heat balance

Calculating the Overall Tower Heat Balance

Knowing the reboiler and condenser duties, an overall heat balance over the tower can be calculated and will provide the preheat required in the feed and consequently its temperature. A calculation to determine the bubble point of the feed at 134 psia (mean inlet pressure) was made. This temperature was 305 °F. Should the calculated temperature be above this, then an enthalpy curve would need to be developed for the feed to determine its actual enthalpy and temperature. The overall heat balance now follows (Table 12).

Table 12 Overall tower heat balance

Enthalpy of the feed is $\frac{21,713,000}{142,659}=152.2\;\mathrm{Btu}/\mathrm{lb}$; from the enthalpy tables, the temperature is found to be 276 °F. The feed tray is based on a preliminary tower temperature profile with a 75 % efficiency for rectifying trays and a 65 % efficiency for the stripping trays (see Fig. 4). Both the temperature profile and the tray efficiencies will be finalized by a suitable computer simulation package or a rigorous tray-to-tray calculation. Note: The calculation techniques and examples given here are good input to computer simulation packages; wherever possible, these packages should be used for final process design.

Fig. 4

Tower temperature profile

Tower Loading and Sizing

Light ends towers all follow the following principles of cross-sectional area sizing and tower height. These dimensions are interrelated by the height required between trays to ensure proper separation of clear liquid from the frothy mixture of the tray inlet fluid. There are many procedures and correlations to determine these dimensions. The following method is just one which will provide a good estimate for tower design. For definitive design, however it is essential that tray manufacturers be consulted and their methodology be used. After all they, the manufacturers, will be required to guarantee the performance of the unit in terms of flooding capacity and tray efficiency. In all cases this has to be a significant consideration, but in case of super fractionating units with over a hundred trays in many cases, this has to be a primary consideration.

There are many light ends tower computer simulation packages in the market, and most of these calculate vapor and liquid traffic in the tower on a tray-to-tray basis. The input for these programs however does require a fairly accurate estimate of these values for easy convergence and subsequent use of the program. The following procedure with the equations used can provide details of tower loading and tray criteria at critical trays in the tower. A linear correlation between these points will then provide a reasonable loading profile over the tower sufficient for good computer input.

Tower Loading and Overall Tower Diameter

The diameter of a fractionating tower is usually based on the vapor loading on two critical trays. These are the tower top tray which will set the diameter for the rectifying section (i.e., the trays above the feed tray) and the tower bottom tray. This lower tray sets the tower diameter for the stripping section of the tower (i.e., the feed tray and the trays below the feed tray). These two sections may have different diameters.

The vapor loading is based on the rate of vapor passing through the tray and the density of the vapor and the liquid on the tray through which the vapor bubbles. This relationship is given by the Brown and Souders equation. There are several forms of this equation which can be used with the appropriate physical tray constants. One of the forms used here is as follows:

$$ {G}_{\mathrm{f}}=K\surd \left({\rho}_{\mathrm{v}}\times \left({\rho}_1\hbox{--} {\rho}_{\mathrm{v}}\right)\right) $$

(5)

where:

• G _f = mass of vapor per sq foot of tray at flood (lb/h · ft²)

• K = a constant based on tray spacing at flood (see chapter “Atmospheric and Vacuum Crude Distillation Units in Petroleum Refineries”)

• ρ _v = density of vapor at tray conditions of temperature and pressure in lb/cuft

• ρ _l = density of liquid at tray conditions in lb/cuft

The area thus determined is the “bubble” area of the tray. Normally trays are designed at 85–90 % of flood. Therefore for good design, the G _f is divided by this percentage to give the actual or design area of the tray. The whole tray is made up of two other areas: that for the downcomers and a waste area which is allocated to calming the liquid leaving the bubble area before entering the downcomer. The relationship of these areas to one another is given in Table 13.

Table 13 Valve and sieve tray characterization

Using the criteria in Table 13 and the value of the bubble area based on the vapor loading, the total tray area and therefore the tower diameter can be determined. This relationship is summed up by the following expression:

$$ {A}_{\mathrm{s}}={A}_{\mathrm{B}}+{A}_{\mathrm{dc}}+{A}_{\mathrm{w}} $$

(6)

where:

• A _s = total tray area

• A _B = bubble area

• A _dc = downcomer area (inlet + outlet)

• A _w = waste or calming zone area (usually 15 % of A _s)

Tray Spacing

The tray spacing used in the initial determination of flood loading needs to be checked. If necessary, the spacing and the calculation will be revised to meet the correct spacing criteria. Usually this first guess at tray spacing is taken as 24″. The following equations are then applied to determine whether this spacing is satisfactory. These equations calculate the pressure drop across the tray in terms of the clear liquid holdup (or height) in the downcomer. This height of liquid must be less than 50 % of the tray spacing for most applications. In the case of a high foaming process, this height must be less than 40 % of the tray spacing. These pressure drop criteria concerning the tray hydraulics and their associated equations now follow:

Clear Liquid Height, h _cl

$$ {h}_{\mathrm{cl}}=0.5\times {\left[{V}_L\div \left({N}_p\times {l}_o\right)\right]}^{2/3} $$

(7)

where:

h _cl = height of clear liquid on tray in inches of hot liquid

V _L = liquid loading in gallons per minute of hot liquid

N _p = number of passes on tray

l _o = length of outlet weir in inches

Effective Dry Tray Pressure Drop, h _ed

The effective dry tray ΔP shall be the greater of the following two expressions:

• (a)
  $$ \Delta {P}_{\mathrm{po}}=1.35\;{t}_{\mathrm{m}}.{\rho}_{\mathrm{m}}/{\rho}_1+\mathrm{K}.\left({V}_{\mathrm{o}}^2\right).{\rho}_{\mathrm{v}}/{\rho}_{\mathrm{l}} $$

  (8)
• (b)
  $$ \Delta {P}_{\mathrm{fo}}={K}_2\left( {V}_{\mathrm{o}}^2\right){\rho}_{\mathrm{v}}/{\rho}_{\mathrm{l}} $$

  (9)

where:

ΔP _po = dry tray ΔP, valve partially open. In inches of hot liquid

ΔP _fo = dry tray ΔP, valve fully open. In inches of hot liquid

t _m = valve thickness in inches (see Table 14)

Table 14 Valve thickness in inches

ρ _m = valve metal density in lb/cuft (see Table 15)

Table 15 Valve metal densities (in lb/cuft)

V _o = vapor velocity through valves in ft/s (= cuft/s/A _o)

Constants K ₁ and K ₂ are given in Table 16.

Table 16 Values of K ₁ and K ₂

In calculating V _o, assume ratio hole to bubble area (A _o/A _B) is 12 %.

Total Tray, ΔP · ht

$$ \mathrm{Total}\ \mathrm{tray}\ \Delta P={h}_{ \mathrm{cl}}+{h}_{\mathrm{ed}} $$

(10)

where:

h_cl = height of clear liquid on tray in inches of hot liquid

h_ed = effective dry pressure drop in inches of hot liquid

Head Loss Under Downcomer, h _ud

$$ {h}_{\mathrm{ud}}=0.06{\left[\mathrm{G}\mathrm{P}\mathrm{M}\div \left(c{L}_{\mathrm{i}}-{N}_p\right)\right]}^2 $$

(11)

where:

• h _ud = head loss under downcomer in inches of hot liquid

• c = constant = 1.5

• L _i = length of inlet weir in inches

Inlet Head in Inches of Hot Liquid, h _i

When there is an inlet weir, use:

$$ {h}_{\mathrm{i}}=0.06{\left[GPM\div \left({N}_p\cdot {L}_{\mathrm{i}}\right)\right]}^{2/3}+{h}_{\mathrm{wi}} $$

(12)

where:

• h _i = tray inlet head in inches of hot liquid

• L _i = length of inlet weir in inches

• h _wi = height of inlet weir in inches

Where there is no inlet weir, then h _i = h_cl.

Downcomer Filling in Inches of Hot Liquid

$$ {L}_{\mathrm{d}}={h}_{\mathrm{i}}+\left({h}_{\mathrm{t}}+{h}_{\mathrm{ud}}\right)\cdot \left[{\rho}_1\div \left({\rho}_1-{\rho}_{\mathrm{V}}\right)\right]+1.0 $$

(13)

where:

L _d = downcomer filling in inches of hot liquid

Example Tower Sizing Calculations

The following is an example of tower sizing and hydraulic analysis. Tray loading data that is used for this example are those calculated for the debutanizer column earlier in this chapter. Thus, all data are at tray conditions of temperature and pressure (Table 17):

Table 17 Example tower sizing basis

Bubble area on top tray at flood:

$$ {G}_{\mathrm{f}}=K\surd \left({\rho}_{\mathrm{v}}\times \left({\rho}_1-{\rho}_{\mathrm{v}}\right)\right) $$

(14)

where:

• G _f = load at Flood in lb/h ft²

• K = 1,110

• ρ _v = 1.11

• ρ _l = 33.37

• G _f = 6,415.8 lb/h ft²

Tower diameter will be designed to 80 % of flood. Then G _A = 5,132.6 lb/h ft²

Bubble section A _B of tray at 80 % flood will be $\frac{57,297}{5,132.6}=11.16$ ft².

Downcomer area (inlet and outlet) A _dc:

Downcomer velocity will be 0.4 ft/s (see Table 13).

Then area of one downcomer will be 0.417 ÷ 0.4 = 1.04 ft².

Total downcomer area A _dc = 2.08 ft².

Waste area of tray A _w will be 15 % (see Table 13).

Total tray area A _s = A _B + A _dc + A _w.

Then for the top section of the tower, the diameter will be:

• A _s – 0.15A _s = 2.08 + 11.6 ft²

• A _s = 15.58 ft² diameter = 4.45 f. say 4.5 ft

Similarly for the stripping side of the tower:

$$ \begin{array}{c}{G}_{\mathrm{f}}=1,110\surd 3.44\times \left(38.9-3.44\right)\\ {}=12,259 \mathrm{lb}/\mathrm{h}\ {\mathrm{f}\mathrm{t}}^2\ \mathrm{and}\ \mathrm{at} 80 \%\ \mathrm{flood}\ {\mathrm{G}}_{\mathrm{a}}=9,807 \mathrm{lb}/\mathrm{h}\ {\mathrm{f}\mathrm{t}}^2\end{array} $$

• Downcomer area = 1.715/0.4 = 4.29 ft² and A _dc = 8.58 ft².

• A _s = 24 ft² and diam is 5.53 ft, say 6 ft.

Tower Hydraulics and Downcomer Filling

Using the pressure drop equations defined earlier, the percentage of downcomer filled by liquid is calculated. This calculation is based on the stripping section of the tower only. A similar one will be completed for the rectifying section. Thus:

Clear Liquid Height, h _cl

$$ {h}_{\mathrm{cl}}=0.5\times {\left[{V}_{\mathrm{L}}\div \left({N}_{\mathrm{p}}\times {L}_{\mathrm{o}}\right)\right]}^{2/3} $$

(7)

• V _L = 768.4 GPM.

• N _p = 2 (liquid loading is relatively high so the option of a two pass tray is used).

• L _o = 58.8 in. Use the correlation given in Appendix 1 of the chapter entitled “Process Equipment for Petroleum Processing” in this handbook.

• Then h _cl = 1.74 in. of hot liquid.

Effective Dry Tray, ΔP

• (a)
  $$ \Delta {P}_{\mathrm{po}} = 1.35{t}_{\mathrm{m}}.{\rho}_{\mathrm{m}}/{\rho}_1+{K}_1 .\left({V_{\mathrm{o}}}^2\right).{\rho}_{\mathrm{v}}/{\rho}_{\mathrm{l}} $$

  (8)
• (b)
  $$ \Delta {P}_{\mathrm{fo}}={K}_2\left({V}_{\mathrm{o}}\right){\rho}_{\mathrm{v}}/{\rho}_{\mathrm{l}} $$

  (9)

Use A _o/A _B = 12 % giving A _o as 1.32 ft².

Valve thickness is 0.05 in., and metal density is 480 lb/cuft.

• K ₁ = 0.2, K ₂ = 0.92

• V _o = 6.6 ft/s

• ΔP _po = 1.603 in. of hot liquid

• ΔP _fo = 3.54 in. of hot liquid

Total Tray, ΔP · h _t

h _t = 1.74 + 3.54 = 5.28 in. of hot liquid

Head Loss Under Downcomer, h _ud

• h _ud = 0.06[GPM ÷ (cL _i · N _p)]2

• c = 3.2 (estimated)

• L _i = 58.8 in.

• h _ud = 0.96 in. of hot liquid

Inlet Weir Head, h _i

There is no inlet weir therefore h _i = h _cl.

Total downcomer filling, L _D

$$ \begin{array}{c}{L}_{\mathrm{D}}=1.74+\left(5.28 + 0.96\right)\left(38.9/35.46\right)+1\\ {}=9.54 \mathrm{inches}\ \mathrm{of}\ \mathrm{hot}\ \mathrm{liquid}.\\ {}=40 \% \mathrm{of}\ \mathrm{tray}\ \mathrm{spacing},\mathrm{which}\ \mathrm{is}\ \mathrm{satisfactory}.\end{array} $$

Checks for Light Ends Tower Operation and Performance

Most light ends towers are very stable in their operation. That is, once they are lined out for an operating requirement under normal unit control, they maintain their stability. When performance falls off, it can be attributed to one of a few reasons. This section looks at some of these reasons and how they can be evaluated and checked. Performance in this case is meant the ability of the unit to make product quality at the prescribed throughput.

Cold Feed

The condition of the feed entering the tower is very important to the tower operation. Ideally the feed should enter the tower at as close to a calculated feed tray temperature as possible. If the feed is well below its bubble point on entering the tower, several trays below the feed tray are taken up for heat transfer before effective mass transfer can begin. This could prevent the specified product separation occurring and tray efficiency in this section of the tower falls off dramatically. Feed condition can be checked by bubble point calculation and a flash calculation.

Hot Feed

This situation is probably the more serious regarding feed condition. If the feed enters at a temperature far above its bubble point, its resulting enthalpy will be such as to reduce the reboiler duty. This will occur automatically as the tower must always be in heat balance. The tower controls will maintain the product quantity and split. However, if the reboiler duty is drastically reduced, insufficient stripper vapors will be available for the stripping function. Poor separation will result.

As a rule of thumb, the stripping vapor to the bottom tray must be at least 70 % mole of the bottom product made. In super fractionation, such as a deisopentanizer, this figure would be at least 80–100 % of bottoms made.

Heat balances as shown will quickly determine the stripping vapor status.

Ideal Feed Condition

Ideally the feed should enter the tower close to the feed tray temperature. Usually then at the inlet pressure, the feed will be in a mixed phase with the vapor portion very close in quantity to the distillate product. As the feed to these units is generally heated by the bottom product heat exchange, the approach temperatures are always a consideration. To maintain good feed conditions, however it is often beneficial to include a separate steam (or hot oil) feed preheater.

Entrainment

A common cause of poor plant performance at high throughout or high reflux rates is liquid entrainment or carryover from tray to tray. Very often in a high load and entrainment situation, the problem is further aggravated by increasing reflux to attempt separation improvement.

A well-designed light ends tower can operate up to about 120 % of allowable flood before substantial carryover occurs. Loading above this figure would result in some degree of entrainment.

Downcomer Backup and Flooding

If tower loadings are increased well above allowable flood point, there is a real danger that downcomers become unable to cope with the liquid load. They would fill, and the tower would be in a state of flood. This will be very apparent with very high, abnormal pressure drops occurring across the tower. Separation by fractionation is not possible under these conditions. Heat input (and feed rate) to the tower must be reduced to bring the unit back to a normal pressure drop.

Low Tower Loading

Most towers have been designed with at least a 50 % turndown ratio for the trays. This means that the trays should operate satisfactorily at 50 % of their loading. Nevertheless tray performance does fall off at these low loadings. Below this turndown ratio, performance, particularly in sieve trays, is drastically reduced. This is almost certain to be due to “weeping” where liquid falls from tray to tray. If the low loads are to be for only a short time due to temporary reduced throughput, tray loading can be increased by increasing reflux. If the low throughput is to continue for an extended period of time, a tray blanking schedule should be considered to reduce the active tray area.

Operating Close to Critical Conditions

Deethanizers in particular operate close to critical pressure at the bottom of the tower. Careful attention should be paid to avoid any pressure surges in this unit. Feed to the unit and reflux streams should be on flow control.

No separation by fractionation can occur at pressures in excess of critical. Very often chilled water is used for overhead condensing to reduce reflux drum pressure but maintaining minimum C₃ loss in the case of deethanizers.