DISTILLATION COLUMN DESIGN

The separation of liquid mixtures into their various components is one of the major operations in the process industries, and distillation, the most widely used method of achieving this end. In processing, the demand for purer products, coupled with the need for greater efficiency, has promoted continued research into the techniques of distillation. In engineering terms, distillation columnshave to be designed with a larger range in capacity than any other types of processing equipment, with single columns 0.310 m in diameter and 375 m in height. Designers are required to achieve the desired product quality at minimum cost and also to provide constant purity of product even though there may be variations in feed composition. A distillation unit should be considered together with its associated control system, and it DISTILLATION COLUMNS A distillation column consists of a cylindrical structure divided into sections bya series of perforated trays which permit the upward flow of vapour. The liquid reflux flows across each tray, over a weir and down a downcomer to the tray below. The vapour rising from the top tray passes to a condenser and then through a reflux drum and a reflux divider, where part is withdrawn as the overhead product D, and theremainder is returned to the top tray as reflux R. The liquid in the base of the column is frequently heated by an external reboiler. The vapour generated in the reboiler is returned to the bottom of thecolumn with a composition y , and enters the bottom tray where it is partially condensedand then revaporised to give vapour of composition y1. This operation of partial condensationof the rising vapour and partial vaporisation of the reflux liquid is repeated on eachtray. Vapour of composition ytfrom the top tray is condensed to give the top productD and the reflux R, both of the same composition yt. The feed stream is introduced onsome intermediate tray where the liquid has approximately the same composition as thefeed. The part of the column above the feed point is known as the rectifying section andthe lower portion is known as the stripping section. The vapour rising from an ideal traywill be in equilibrium with the liquid leaving, although in practice a smaller degree ofenrichment will occur.

Methods and experimental data

This design is required to concentrate ethanol from a waterethanol mixture. The pressure of the column will be designed at atmospheric pressure and the feed to the column will be saturated liquid with 20% w/w ethanol with a total mass flow rate of 20kg/sec. The distillate product must be at least 95% w/w ethanol, with no more than 1% w/w ethanol at the bottom.However distilling a water-ethanol mixture to produce such high purity ethanol has deemed difficult because the ethanol forms an azeotropewhen mixed with water. To overcome this problem, the azeotrope can be eliminatedby the addition of diaminoethanal at concentrations between 2.26 to 4.6 kmol/m3(Kumar, S. 2010). For this specific design, the distillation column is based on a binary system, and the chosen concentration of diaminoethanal is 4.6 kmol/m3. The performance of a distillation column depends on its sizing, therefore the following systems werestudied, and specifications of each of them were obtained: Column Insulation of the column

Condenser Reboiler Pump

The results were analysed and the optimization parameters of the systems wereestablished. Column Reflux ratio and number of trays To determine the number of trays and the reflux ratio (R), the Mcabe-Thiele graphical
approach was used, which assumes constant molar overflow, which applies that:

constant molar heat of vaporisation no heat losses no heat of mixing

The molar concentrations of ethanol (Kumar, S. 2010), at 4.6 kmol/m3diaminoethanal, were used to plot a vapor-liquid equilibrium curve, and a 45 line was added as follows:

The operating line for getting Rmin was drawn, intersecting zf, xd (pure liquid) and the y axis. The intercept with the y axis is equal to: xDRmin+1 Rmin = 1.766 was determined from this equation, as our actual reflux ratio dependson the minimum reflux ratio. The actual reflux ratio (R) was obtained by: R=fRmin Where f is a factor that was varied from 1.1 to 1.5 in intervals of 0.1 for further analysis. The following plots and values were obtained: f R Theoretical Actual number of

number of trays

trays (Theoretical number of trays / 0.6) 19 15 14 14 12

1.1 1.2 1.3 1.4 1.5

1.942 2.119 2.295 2.472 2.649

11 9 8 8 7

The value of R used for designing the distillation column was 2.295, giving 14 trays. The optimization of these values will be discussed later. Material Balance The flow rates (F, D, B,Vn, Vm, Ln, Lm) were determined by doing an overall material balance around the distillation column. The following equations were used to obtain B and D: F=B+D zfF=xBB+xDD The equations used to obtain Vn and Ln were: R=LnD Vn=D+Ln As there is no vapour in the feed: Vn=Vm Lm=B+Vm Giving the following vales Stream Feed Distillate Bottom Top Vapour Top Liquid Bottom Vapour Bottom Liquid Stream Label F D B Vn Ln Vm Lm Molar flow rate (mol/s) 975.8454106 94.71682598 881.1285846 299.1157365 204.3989105 299.1157365 1180.244321

Height of the column The temperatures at the top and the bottom need to be determined. At the bottom, xB=0.0039, T=98.85C which were obtained from the data table(Kumar, S.

2010). The temperature at the top, xD=0.88 was obtained by interpolation giving a T=81.83C. The densities of the liquids, at the top and the bottom, were determinedusing Perrys handbook data and using the following equation: 1l=mee+mww Giving values ofltop=742.25kg/m3 and lbottom=853.52kg/m3 It was assumed that the vapors behave as ideal gases. The densities at the top and at the bottom were obtained by using the average molecular weight as follows: MWav=yBeMWe+yBwMWw Giving values ofMWavtop=44.09g/s and MWavbottom=20.04g/s. These MW were used in: p MW=RT To get vtop=1.51kg/m3 and vbottom=0.65kg/m3. The densities were used to get the values of the maximum vapor velocity (Um): U=Klv Where K=0.27L-0.17L2-0.047, and using a plate spacing (L) of 0.5m. The values obtained wereUmtop=1 m/s and Umbottom=1.63 m/s. The actual vapor velocities (U) were obtained by multiplying the Um by 80%. The results obtained were Utop=0.8 m/s and Ubottom=1.31 m/s. The volumetric flowrate (Q.) was calculated by: Q.=mv Obtaining Q.top=8.8m3/s and Q.bottom=9.16m3/s. The diameter (d) of the column was calculated by: d=4Q.U dtop=3.72m and dbottom=2.99m were obtained. The largest d was used for further calculations, as a larger d gives a lower vapor velocity and prevents splashing. The equation used to calculate the height of the distillation column was: h=2d4+2+3+n of trays0.5 The height without the skirt obtained was 11.36m, and by adding the skirt we obtained an actual d of 13.43m. Insulation The heat loss (Qloss) was determined without insulation, and then the insulation was added. Iteration was conducted to obtain the optimum thickness of the insulation. Heat loss without insulation The resistances were determined for the wall of the distillation column and for the air using the following: R=1UA=t(Am cylinder+hemisphere)+1A0 cylinder+A0 hemisphereh0 Am=Ao cylinder+Ao hemisphere-Ai cylinder+Ai hemisphereln (Ao cylinder+Ao hemisphereAi cylinder+Ai hemisphere) Acylinder=2rH Ahemisphere=3r28

The R obtained was 0.00058 K/W. For calculating the Qloss the To=5C (winter) was assumed and Ti=98.85C was used as it is the temperature at the bottom of the distillation column. The Qloss was calculated from the equation: Qloss=TR The Qloss obtained was 161.67 kW. Heat loss with insulation The resistances were determined for the wall of the distillation column, for the wall of the insulation and for the air, different from the one without insulation due to an increase in the area by the insulation, using the following: R=1UA=t(Am cylinder w+hemisphere w)+t(Am cylinder i+hemisphere i)+1A0 cylinder+A0 hemisphereh0 The same temperatures as the Qloss without insulation were used to obtain a Qloss with insulation at different thickness values. Thickness (mm) Qloss (kW) Temperature outside the distillation column with insulation (C) 0.025 28.11 20.9 0.05 15.5 13.6 0.075 10.74 10.9 We will use the 0.075 thickness because is the one with the less heat loss. Even though it will be an expensive fixed cost, it is a one time investment that will not be that high when comparing it with the operating cost throughout the years. Condenser For designing the condenser, the duty (Q) needs to be calculated as follows:
Q=-mmixtureHev mixture

mmixture=methanol+mwater Where mmixture=12.76 kg/s and Hev mixture =270 kcal/kg was obtained from Perrys handbook. The Q lost was 14434.35 kJ/s. The Q was used in: Q=mcwCpcwT To obtain mcw=230.37 kg/s where we assumed that T=15C, and Cpcw was obtained from the Perrys handbook. The Q.cw was obtained from: Q.cw=mcwcw

Q.cw=0.232 m3/s where the cw=992.21 kg/m3 was obtained from Perrys handbook at 27.75C. To calculate the theoretical number of tubes inside the condenser, the cross sectional area was calculated first: Ac=Q.cwU U was assumed to be 1 m/s, and the Ac found was 0.232 m2. The theoretical number of tubes equation used was: n=4Acdi2 The theoretical number of tubes obtained was 1810. The overall heat transfer coefficient (U0) was calculated from: 1U0=D0Di1hi+D0DiFi+D02lnD0Di+F0+1h0 Where hi=0.023[Ud]0.8[Cpk]0.4d Finding a value of U0=1.27 kW/Km2. The two fluids entering the condenser are in counter flow .The cooling water is entering the condenser at 20C and leaving at 35C, and the water and ethanol mixtures enters and leaves at 81.3C. With this in mind we calculated the Tm by:
Tm=T0-TLlnT0TL

Obtaining a value for

Tm

=53.44C. A0=QU0TmF

The U0 and Tm were used to calculate the A0, as it is shown: Where F is the correction factor and it is equal to 1. The A0 found was 212.64 m. From this value we can obtain the length of the tubes inside the condenser by: L=A0nD0 The value obtained was L=2.34, and as we can only choose between 2.44 m and 3.66 m, the 2.44 m tube was selected and the actual U=1.04 m/s was calculated. Reboiler For obtaining the reboiler duty and the volumetric flow rate of steam produced, the following steps were followed. The duty (Q) was calculated with the same formula as the one for the condenser. Where in this case mmixture=21.37 kg/s and Hev mixture =100 kcal/kg was obtained from Perrys handbook, and a value of Q = 8950.79 kJ/s. By dividing the mass flowrate of vapor entering the column we can obtain the volumetric flow rate of steam produced by the reboiler, obtaining a value of 9.22 m3/s. Pump The pressure drop of the pump was calculated my the following method. Method 1 The optimum diameter of the pipe was calculated by: di=0.8(Q.)0.5 Being Q the volumetric flowrate of Vn, giving a value of di=108 mm. As there are no pipes of this diameter available, the selected diameter was 110 mm.
.

The velocity accross the pipe was obtained next. It is said that the optimum velocity of a at a density of 800 kg/m3 should be around 2.5 m/s. Also, from literature, the accepted pipe velocity for a water-ethanol mixture is in the range of 1.5 to 3 m/s. Our velocity was calculated from: Ac=Q.cwU Where: Ac=di24 Giving us a U of 1.98 m/s Reynolds numer was calculated using the following data: =742.25 kg/m3 and =5*10-4. The material chosen for the pipes was iron cash, with an absolute roughness of 0.26, and a relative E=2.4. Using Moody Diagram we extrapolated our friction fanny factor f = 0.0028. The total length of the pipeline was assumed to be around 24 m. The friction loss in the pipeline was calculated by: Pf=8fLU22di The value obtained was Pf=3000 Pa, which is an accepted value. Literature says it should be less than 0.1 atm. The maximum difference in elevation (z1-z2) was around 12 m. And the pressure difference (p1-p2) was assumed to be around 0 Pa. The energy balance was conducted by: zg-p-Pf-W=0 W=-121.6 J/kg was obtained. And using this work we get a pump power of 30 kW.

D= molar flow rate of the distillate B = the molar flow rate of the bottom F= feed flow rate Xd= mole fraction of liquid ethanol y= mole fraction of vapour ethanol xb= mole fraction of ethanol at the bottom xf= mole fraction of ethanol in the feed. We take the mole fraction of ethanol as the vapour is always more richer in the volatile component The flow rates of liquid and vapour leaving a tray are labelled asVn and Ln respectively. Were we are assuming that

There are no heat losses from the column The difference in temperature between adjacent trays is negligible The latent heat of the two components are equal

in other words for every mole of vapour condensed, one mole of vapour boils off and thus for every mole of liquid flowing to a tray, one more flows liquid flows from that tray( they are in equilibrium) The vapour from the condenser is completely condensed so that the resulting liquid has the same compostion. This liquid is split to produce the top product and to provide liquid for the top tray.Thus theses three streams, the vapour leaving the top tray and the top product and the reflux all have the same composition Xd.

As the condenser at the top of the column is a total condenser the vapour flow rate is equal to the sum of the reflux and the distillate flow rate: Vn=D+Ln At the feed tray the vapour and liquids flow rates in this section are denotes as Vm and Lm this section is called the stripping section. Where Vm = Vn And Lm= Ln+F Determining the Number Of Trays Now that we know xd = 0.88, xb= 0.004 and zf= 0.089 we can determine the Reflux ratio: R=Ln/D The reflux ratio is used to calculate the number of trays in the column by applying the Mcabe-Thiele graphical approach which assumes constant molar overflow, which applies that:

constant molar heat of vaporisation, no heat losses no heat of mixing, This allows us to apply two operating lines above the feed (enrichment section): Which we can relate to Reflux by substituting Ln and Vn

by the relations Ln= RD and Vn=(1+R)D Giving

And below the feed (stripping section)

Initially we determine minimum number of trays, which will effect component cost. However, less trays means a higher reflux ratio and so a larger condenser and more reboiling. As a result operation cost will increase. Therefore we need to optimise the number of trays to the heat exchanger. At first we need to determine the minimum reflux ratio, as our actual reflux ratio will be chosen in relation to the minimum reflux ratio.The minimum reflux conditions represent the theoretical opposite of total reflux and infinite number of ideal separation stages. IN this case the intersection of the operating line lies on the equilibrium line. Thus the distance between the equilibrium curve and ioperating line is at its minimum, the stepping triangles become very small, and therefore there s no gap between equilibrium cuvre and the intersection point. Sou you cannot step past the feed point.

To determine the number of trays we will apply the McCabe-Thiele graphical approach. It assumes constant molar overflow and this implies that:

constant molar heat of vaporisation, no heat losses, and no heat of mixing, l Which lead to a constant molar vapour flow and a constant molar reflux
Our VLE plot was taken from experiment date shown below (Kumar, K. 2010). Experimental data concentration of diaminoethanal = 4.16 Kmol m^-3

Mention m q= 1 as its saturated liquid feed

Results and Discussion:

Optimization of reflux ratio Quantitative analysis: First we would want to explain the true physical meaning of the reflux ratio R, and its inverse relation to the number of trays in the distillation column. The reflux ratio is fraction of condensed ethanol vapour sent back to the column to achieve total separation of the two components.

Therefore, by increasing the reflux ratio, we are increasing the purity of our distillate, hence improving our process. Another factor affecting the quality of the separation is the number of trays used. Increasing the number of trays, intuitively, will result in a better separation. So for each reflux ratio, there exist a specific number of plates that gives total separation. We must therefore evaluate the trade-off between reflux ratio R, and the number of trays N. At the start, our R was unknown, so we needed to calculate the Rmin and then get our reflux patio by iteration. The R minimum, describing a return of a minimum amount of liquid to the column, correspond to an infinite number of trays (separation is impossible at this point). This point is referred to as the pinch point. By increasing R, the number of trays required start decreasing until reaching the conditions of Total reflux, which corresponds to the minimum number of trays needed, and an infinite reflux ratio. So in our design, our reflux ratio R should correspond to the optimized, or the most economical, for which the total cost, will be the least. There are two components of total cost: fixed capital costs fc(such as the cost of purchasing the distillation column, reboiler, condenser) and the operating cost fo (costs associated to operating the plant; such as condenser cooling water, vapour steam). At the minimum reflux ration Rmin , the operating cost is at minimum because we are handling a minimum amount of liquid. So this is the stage of the minimum reboiler heat duty, and the minimum condenser cooling capacity. But the fixed capital cost is at maximum due to the infinite number of trays required. Fixed capital cost:
-

Number of trays Size of distillation column

Size of reboiler Size of condenser Size of the pump Piping and instrumentation (tubes) Insulation Maintenance operating labor supervision plant overheads and laboratory facilities taxes, and insurance and effluent treatment

Operating capital costs

-

Steam Cooling water Electricity NB: raw material is also included in the operating costs nut it does not vary the change in R.

As R start increasing, larger equipments are required, and the duties of the reboiler and the condenser start increasing gradually with the increase of the overhead product back to the column as reflux and reboiling all bottom products. In the first approach to analyze the trend of the fixed capital costs variation with the increase R, one would assume a linear continuous decrease of the fixed cost as R increase due to the decrease in the number of trays. Now, if we take a deeper look at our fixed capital cost, we can see

that our fc is a function of three variables that react independently and correlates differently with fc . fc = fc (x,y,z)
-

x: piping maintenance, operating labor, supervision, plant overheads and laboratory facilities, taxes, and insurance and effluent treatment

y: number of trays z: distillation column, insulation, pump, reboiler, condenser

x varies minimally as R increases, so we can assume it to be constant and has no effect on our economical analysis. dx 0 At minimum R, we have an infinite number of trays and a minimum duty on the heat exchanger. So the variation of fc as R increases is set by the effect of component y since it is clear that dy >>> dz As R continue increasing, y decreases but our z start increasing. At small values of R, fc starts decreasing due to the decrease in y (since the trend is governed by the effect of y). At high values of R, although y is at a minimum, we would need larger heat exchangers, larger pumps and bigger insulation to account for the larger amount of liquid, and therefore heat loss, we are dealing with. These cost much more than tray does. Hence, dz >>> dy. So our fc starts increasing again governed by the increase of z. the effect of the cost of trays is negligible. Therefore, fc falls into a minimum then rise again to infinity. As for the operating cost fo , it will continue to increase with increasing reflux ratio.

Quantitative analysis:

First we will plot the variation of the number of trays versus the reflux ratio, and we extrapolate.
R Actual num of trays ber 1.9426 19 2.119 15 2.295 14 2.472 14 2.649 12

At R= Rmin the numbers of trays y = As R

30 28 26 24 22 20 18 16 14

,y

Rmin

s y a t f o r e b m u N

12 10 8 6 4 2 0 1.6

R in m

1.8

2.2

R efluxra R tio

2.4

2.6

2.8

Next, we will analyze the effect of the varying the reflux ratio R on the operating cost fo . We now that the highest operating cost, is that of the heat losses. Heat loss through insulation (10.74kW) represent 0.4% from the total heat loss through the distillation column (heat loss through heat exchanger is 23385 kW). So our operating cost of will be a function of the duty of the condenser Qc and duty of the reboiler Qr).

Relationship between condenser duties and reflux ratio The duty of the condenser: Qc= -mmixtureHev mixture mmixture= -Vn MW Vn=D+Ln
R=LnD Vn=D+RD=D(R+1) Qc=-DR+1MWHev mix

We have D,Mw, Hev mix Constant SoQc R+1, Taking a ratio of 2Q To R1 and R2 Qc D(R1+1) Qc D(R2+1) Q1cQ2c=(R2+1)(R2+1) We can say that the duty of the condenser is a function of one variable, R.

Duty on the reboiler:

QR= mmixtureHev mixture Mmixture=LmMW QR=LmMWHev mixture Lm=FL+Ln=FL+RD QR=F+RDMWHev mixture =FLMW Hev mixture+DMW Hev mixtureR Let: A=FLMWHev mixture andB=DMWHev mixture y=a+bx Now we will take the ratio of Q1c to Q2c Q1RQ2R=F+R1DMWRHev reboilerF+R2DMWRHev reboiler Assume MWR constant (assume pure water at the bottom) Q1RQ2R=F+R1DF+R2D

Now lets get a relation for Q1+Q (Duty across heat exchangers) QR+QC=-mC mixHev condenser+mR mixHev reboiler=-DR+1MWCHev condenser+(F+RD)MWRHev reboiler Assume: MWC=MWE MWR=MWW Set the constants A=-D MWE Hev ethanol B=F MWR Hev reboiler C=D MWR Hev mixture QR+QC=A+AR+B+CR=RA+C+(A+B) Set: A'=A+B C'=A+C QR+QC=A'+C'R

So we can say that Q is a constant for a specific reflux, and that this is a linear relationship between Reflux rate (R) and the duties across the heat exchangers. We will plot the variation of Qc+Qr versus the reflux ratio.

So now we have both of the operating cost, and the fixed capital cost trend. Our total cost fT, which is the sum of the two costs, fo that increases linearly with r, and fc and passes through a minimum, must therefore pass through a minimum. The reflux ratio at this minimum total cost, is our optimum reflux ratio. Since we do not have enough data for an economical analysis, we will do a rough estimation of the R optimum from data gathered.
R 1.1Rmin 1.2Rmin 1.3Rmin 1.4Rmin 1.5Rmin R Num of trays ber 1.9426 19 2.119 15 2.295 14 2.472 14 2.649 12 Qr+ Qc 3679.203737 4358.76845 5036.792201 5718.668359 6400.544517

After finding our Rmin, a rule of thumb is that the optimum R is between 1.1Rmin and 1.5Rmin. It is clearly not 1.1Rmin because of the high number of trays which will lead to the high fixed cost. Also, 1.4Rmin is neglected because with the same number of trays, 1.3Rmin has a lower reflux ratio and hence a smaller operating cost. 1.5Rmin has a real high reflux ratio, and can be therefore neglected. Now we are left with 1.2Rmin and 1.3Rmin. Our optimized reflux ratio will correspond to the minimum on the fixed capital cost curve. At a minimum, the change in the capital cost to change in reflux ratio should be zero. Since the capital cost is governed by the number of trays at this stage, we can say that dfcdR = 0 at R=1.3Rmin since a variation in R at this point will give the same number of trays. Hence we can conclude that our optimum reflux ratio is R= 1.3Rmin = 2.295 Our total cost curve versus our reflux ratio, if given all the data, is fitted to the following curve: