Fractionation

- EgwiEbfimStage
- Relative Vda3_ility
Perifosrnaa;ee Specifications

- Recovery

= Design ons side rations

- Pressure
- Reflux Ratio and Number of Stages
- Tray Effxeiency
- Feed Location and Enhlpy
s fnternds
1 Most distiltation is modeled using i'equiXibrium

u A pmponent has a vapor liquid equilibrium K

' value that is defined as the m o k ratio of its
vapor concentration to its liquid concentration
when'these phases are in equilibsium.
Equilibm'um - Rehtive Volatility

m Alpha (relative volatiEQ) is a measure of the

intrinsic difficulty in using fractionation to
separak two components
n It is the ratio of the vapor liquid eqaibriam
K values for two components
LK = Light Key Component:
HK =Heavy Key Component

Alpha Variation
 AlTpha Vanktion

m A knowledge of the alpha value behavior is m

important piece of infomation for designing
d&tillation colunrzks
 Equilibrium - Ideal Systems
Pmsuse c u n s m

Equilibfium - Non Ideal Systems

Pressure Cunstunt

w e Wiation from X W t y :
e.g. M m m wiing am?t10pe
 How does one measwe performiance of a
fractionator?

Let us define &e performance goals.

Perfomance Goals
(cmSmed)

-
h r i t v andlor Recovem Fur this typical two a t fractionator,
&ere is a Light key cumpownt pu&y spedied for the
&stitlate and a bmvy key cornponeat pwity specified for tixe
bot&m. An afternative god wm#dhave the #ightkey purity
speaed a d the Eght key recovery spx%ed.
-
Capacitv N o d y &e feed rate is spe&ed. Sometimes the
bottom or disljuak fiow is specified instead of feed-
 Performance Goals
{continued)

-
Parity and/or Recoverv For our example reformer
tkbutaizer, the overhead l i e d produet caa contain,0-5
mol%CS+, while fie bottoms product can contain 0.5

-
Capacitv For our example, the design feedrakis46,45Q
B3PSD (J -80 h%MSCFDof everhad vapor product, 1530
BX%D of overhead liqaid prodmet, and 44,596 3PSD of
bogttonrs producg with the prodact purities specified above)

P e d o m n e e Goals
{co-k~lied)

indepeadent tyariabiesa e available tafhe process

%%%at
en- to t o e operational c h q e s ?

Fied bv Ilesim Available for Control

Reflux (Partial) Reflax, 50-125% design
Overhead Coanpositio~
Bottoaa Coxnposition

Overhead Composition
Recovery of. Light Keg:

Can one show graphically the reIatio11shipof the operating

 Pe$unnance Curve

Show gqhicaffy the regaeonship of tfie operating

variab1es

L'iht Key In Bottoms -Y-

e Recovery Of Liht Key

optimizing

What o p d o n d process variabges can one

c;bange far a typical fractionator?

r Pressure
r Reflux
r Feed T h e d Co~ditions
What operational process variables can one change
fur a typiad fractionator?

What are pros and eons of raisingpressare?

optimidng
fcu~&aued)

What operational process variables one &age

for a typical fmctbnator?

What are the pros and cons of raisin.^ W F @ 4 '

deign feed rate?

Against (Con)

Favorable (Pro)
What operational process variables can one change
for a typical f actionator?

What are Uxe pros and cons of increasing the feed

thermal conditions?

Perfomrzance Goals
H e a Hydrocarbon
~ MixIures

Separation is defined as the spread between %he95 vol-O/o or

end point of a labomtory d M i h t i o of
~ the lighter cat vs the 5
vol-% or initial b o i pint of t.he heavier cut,

No&y the feed rateis -&.

 Pe~onnanceGoals
(continued)

%%atindependent variables are available to the process

engineer to make operational changes for heavy
hydrocarbon columns?

Available for Control

Feed T b m a l Condition
Sidecnt Draw Race
Pumpap.~uM3: Heat Removal
Overhead Refla Rate

Gan one show graphicallythe relatioltlsbipof the operating

variables? Optimization of this qpe ef distillation
opesagon is complex and will not be d i s c w e d at this time.

Perfomafzce Goals
(copzcinaed)

IR order to meet these performance goals, the

process design sf the frac~onatormust have (for
&e design feed) the proper combbation of:

r Theoretical Stages
Proper Feed Point
r Feed T h e d Conditions
 In addition, &e mechhanicaf design or hardware
ef the fractionator must function in the

m Tray Efficiency
m Mass Tragsfer Devices
r Coadenser
I Reboiler
i Fractionator N&zles & InternaLs

.! Insmmmtatjon

McCabe-Thiele Diagram

In order to meet these performance go& the process

design of the fra&ena&r must have (for the design
F A ) the proper combination oE

r Theoretical Stages
r Proper Feed Point
r Feed Thermal Conditions

These variables caxl be shown together graphidly on

a McCabe-Thiefe Diagram for a typical fractionator.
 Me Cabe-Thiele Diagram
(co2tti;1~~ued)
Slcetch of Cofumn to 1\,Iodelon McCabe-ThieXeDiagram

Rectifying Section

Me Cabe-llrhiele Diagram
(continued)

n?of fradon light key in vapor

mol fra!&i~n EgM key i~ liquid
mol fraction h v y key in vapor
aaog fraction heavy key in jliqaid
m& per hour vapor-rectifying, strippink
mofs per b u r Iiquid-rmying, stripping
ml fractionlight key ixr feed
 McCabe- Thiele D_iaqam

Siganificax~ceof Points on ,McCabe-Tfrieie Diagram

Colamn Operation
Show graphically the relatiomhipof &e operating variables

McCabe-W3e Diagram
Show p p h i d y the rdationskaiip of min and max RID

Column Operation
(co&*naet%)

What happens i&

Feed tray Too high? Too tew?

Feed Q goes From 0.5 to 1.0? From 0.5 dO?

 Cobmn Operation
(cont%mtled)
Show ppMca:Hy the relationship of &e operating variables

Column Operation
(cmttinued)
 Column Oper&*on

0 goes from 0.5 to 1.0

Column Operation
(conz%atce.d)
Show gmpbically the relatbnship of the operating variables

0 goes from 0 5 to 1.0

 Column Operation
(corztirzued)
f n crrder to meet these performance goah, the process design of
the fractionator must have (for the design feed) the proper
eomf>imtio~ o,f:

i Ream
Theoretic& Stages
Proper Feed Point
Feed T h e d Cond%ions

These variables ean be shown graphically on a McCabe-Thiek

Diagram. Any questions? When is a McCabe-Thiele valid for
design?

These variables can be sagved analytidy with a rigorom

mathematical model. How can we do this?

Computer Cakalmions

Must compucter software win sigoroesly mode3 a

hetionation column by writhg a gmup of
equations fw eaeh theoretical tray.

These equations describe the steady state operation

Heat Batanee
Mass Balance
Vaporlfiquid Equifibrium
Designing a Column

a Define Feed
1 Define f mduct Specification
r Set C d u m Pressure
r Optimize Column Design
m Tray 1EM"tcieixcy
r Mass Transfer Devices

r Composition
8 How Itate
m Temprrrtrrre
s Pressure
r Enthdpy
 Define Product Specifcations

m Receiver kmperature

3 h d u d puri&iesand reeevesies

r Zero purrily spec is not acceptable

r Does eulstomer have.a good definition of the

desired purities and recoveries

m Consul$with sp&&t andproject manager

for streams internal b d t or complex

m Determine the highest purities &at the

c d m ever fias lo produce
,
 Set Column Pressure

r Maximize alpha value

t Wnimize wlum cost
, Keep flare materid out of overhead
r Totally condense overhead prodwcts
r Prevent need for net gas compression

Set Column:Pressure
{co~ti~tued)

jYfiniulrkze net overhead vapw

r Use colndenser as heat marce
r Use boa~nxsas tLot oil
= Limit bo- temperature
- PeIymelization
- Approach to crit5~aI
Feed Preheat Efficciency

Preheater Daty, RrlBtuh

Muti-Component Distillation

u Shortcut Methods for Sizing

- Fenslre equagon
- Underwood ecpfha
- Giflilandgraph
m ExactMethods
- Simultaneous equations
- Top downlbogom up
- Trays are always specified

m Apptorrinxate Design M e t b e
- KTemr $01absorbers
- and strippers
- Naphtha fractionation

rJndemmd - Miaimuxn reflux

Fenske - MrIdxnm tray
 Underwood Method
whimurn Reflux)
Underwood's Method Assumptiom:
Constunt MokZ Overflow m d ConstaPzt a

X - Mo1 fraction in total stream

Q> - l i u a d e m d parameter
Q - Feed thermal wndithns

N - Total number of components

AB possibfe roots ofofquatbn (a) fie betxvem Zbe a'sof the feed

Substimien of these roots into (b) yields @A3)m

Mi~:im;z.cm
Reflax
Sample Problem

- - MokM[r
Ovhd Battom

Bubble Point Feed

 Minimum Regtax
Underwood

(6 x0.4) (2x0.2) (2x0.~) (0.5x0.2)

4"-
6 ( 2 - (I-))' (0.5-4)

1 5 9 <2 (9 between 1 and lightiheavy key alpha)

4 = 1.2267 By Traii and Error
 Fenkse Equation
(Mi~imumTrays)

n = minimm trays
ratio
I=- , of'ligh2: and heavy key mazerial in distillate

r, = r&ioof light and heavy key materia1 in bottoms

Fenkse Equation
(Sample Bobhm)
 Component NBB Feed Ovhd Bottoms
oc -MdSM[r-

36 100 ZOO
69 140 340

80 XU4 204
" 97 200 200

Refomate Splitter
Overhead .to SuEJ"oEdtneBottoms to Zsomm Parex

Component MEP Feed Ovhd &&.oms

"F -Mom-
121 559 553.4 5.6
126 150 149

136 329 1.9 127.1

 Refonnate Splitter
Overhead to Sulfolane Bu~umsto Isomar Parex

Conrponent NBP Feed Ovhd Bottoms

OF -MoIsMLr-

151
2'151 3264.1 5969

m Toluene in bodtoms
Recover 1.0 m d %

E EB in overhead
Recover X.5 MIQI 9%
 Operating Chart

+-Reflw 100%

0 0.001 0.002 0.003 0.004 0.005 0-006 j

EB OvZld

Estimating Component Dislrz3ution

my VaEEcable?

r To Esthmge COX- Material Balance

r To Start a Tray-to-Tray Calculation

 Estirn&*ng Component Dism*hution
Approximate Method

m Conzaponen&lighter than light key - all to

i Components heavier than heavy key - all to

r Only valid 2f not key components and a's are

considerably different than a of keys
m No distributed. components

rdfs = fJ19.6A?.I j/f3.OM.8} = 346011

n = Log {rdrBj/Log (alpha) = 4.539L I73 = 26.21
tag(rdrs) = n * h g (alpha)
4 = fyDlrBrjFIDH/fBH+ (rDI/r~iif
 Tray Eflciency
(continued)

Real Life SeIdorn Me& Criteria for Theomtical Stage

I, V, "A" ia equif%bdmwith inlet liquid
2. V,,t CCBf'i,equilibrium wi& ouffet tiquid

Tray EjJkiency
fcohed)

Theoretical Sgages
R d Stages =
Tray E@deftcj

Tray efficiency obtained from:

I Experience
I Judgment
I Rules of Thumb
t Cddation Methods
 Empirical Predictions
Tray Eflciency vs. Alpha Alone

AIpha iI Tray Efficiency 1

1.2 I 90 i

Mass Transfer Devices

Sieve Trajr
- UOPdefanXt
- 2 to X operating range
- Check customer preference and desired and
expected -down I

r Vdve Tray
- Co& a b u t 20%more than sieve tray
- 5 fo 1 operating range
 Mass Transfer Devices
(continued)

- Cost may be 5 times sieve tray

, - P r m r e d r ~ pmay be VS that of sieve tray
m Bubble Cap Trays
- Cost m y be 3 times vdve tray
- &is nri*m figaid weepingheakage

- ~[igttestcraPadity
- ICowestefficiency

Mass Traasfer Devices

(con$inu:ed)

T m PrubEems

- Vapor or jet Bood (massive entritinxnent)

- Liquid or downcomer backap fZood

- Enwfficient liquid
- Bx&ve boaup
r Damaged Trays
 Mass Transfer Devices
(continued)

pack in^ ]Probiem

Support Grid
- Migration of packing
m Hold Down Grid
- Migration of paching
n Vapor Distribution
r Liquid nj&ribution

Other Thoaghts

R Treme~3.dousQuantity of Material on
Wistiulation (books, causes, etc)
= Mmy shortcug methsds we have not
discussed here (Kremser, Gmand, etc) &at ,
provide fonnufas for estimating lmfnixnurn
refix, minimum stages
m MeCabe-Thiele and other graphical!methods
can be useful for providing a '+@ictoriat"
insight into a design or operatio4 situation!
 B Reformate Splitter

, 1m Determine Number of Trays

r Determine an Optimal Feed Location

 Reformate Splitter
Overhead to SuEfotczne Bottoms to Isontar Parex

Component: NBP
Of: -
Feed Ovhd Bottoms
i&lo$smr-
111 359 553.4 5.6
126 150 149

136 129 1.9 127.1

Refomate Splitter
Overhead to Sulfolane Bottoms to Isomr Parex

Component NBP Feed Ovhd XW&ms

"F -Mom-

--
2151 1264.1
-
8959
 Specifications

Toiuerxe in bottoms
Recover 1.0 mol %

r EB in overhead
Recover 1.5 mo1 %

r Assume 30 theoretical stages

r Set feed stage to be at Tray 15

r Repeat with increments of 2 up through 50

Track the change in Rebvifer Dub

 Economic Evuluu~on

Assume $10,800 for every 2 theoretical trays

Assume Utility Cost is $15O/NIMKcal,

.4ssume 8,000 operating hours annually

Assame 1 year payoaf

(Optimal point is where incremental energy
costlsavhgs .equals the mpiM cost of
addingLwbtracti3i~g2 tiheureticaI trays)

Assume 70 % .tray efficiency

Results - Ntllmber of Trays

 Feed Tray Location

r Feed tray currenkly in middle of column

m Run simufaGon with feed tray located

between Trays 14 and 21
 Feed Tray Location

m "Optimalr'iocation appears to be Tray 16

y Therefore actual Tray 23 (assurmfng 70%

efFrciency) is feed location in a 54 tray column

However,. ..
. .

Final Thoughts

a 33d n a b r .of-ys and feed bation shtoufd

be determhed by using engineering expefience
not just s-om.
m Need to accomE for operating flexibility and
cornpsition chmges.
E Rely on sound enghfx!ringjudgment,
a 'CRules of thumb" (Such as CSDon'tadjust if
there is less than 2%change in Reboiler Duty").