Ef

Ga

b

c

ARRA

KDMF

1

vtp2kmafilP

ofga

iD

0h

Industrial Crops and Products 49 (2013) 437– 444

Contents lists available at SciVerse ScienceDirect

Industrial Crops and Products

journa l h om epage: www.elsev ier .com/ locate / indcrop

conomic feasibility assessment of one and two stages dryractionation of palm kernel oil

regory F.L. Koaya,b, Teong-Guan Chuaha,c,∗, Thomas S.Y. Choonga

Department of Chemical and Environmental Engineering, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor D.E., MalaysiaAdvanced Oleochemical Technology Division, Malaysian Palm Oil Board, 43650 Bandar Baru Bangi, Selangor D.E., MalaysiaInstitute of Tropical Forestry and Forest Products, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor D.E., Malaysia

a r t i c l e i n f o

rticle history:eceived 10 January 2013eceived in revised form 12 May 2013ccepted 19 May 2013

eywords:ry fractionationathematical model

easibility assessment

a b s t r a c t

Palm kernel oil can be fractionated into stearin and olein fractions. Traditionally, industrial players onlyfractionate the softer stearin fraction and harden it through complete hydrogenation. One stage dryfractionation yielded 37% of softer or 24% of harder stearin fraction. Two stages dry fractionation yielded24% harder and 11% softer stearin fractions. Even though two stages dry fractionation salvaged up to14% of the softer stearin fraction from the normally discounted palm kernel olein fraction, the doublefractionation process was not always commercially feasible. Deciphering actual price statistics revealedthat on a per metric tonne basis, the refining premium ranged from MYR90 to 250. Palm kernel oleindiscount was as high as 9% but could command premium over crude PKO by up to 2.5%. Due to the

intrinsic relationship, the corresponding palm kernel stearin breakeven value could range from discountof MYR90 to premium of MYR1260 over crude PKO. The actual 60 months average market premium forpalm kernel stearin stood at MYR870, promising an average MYR430 returns to refiners and processors.By incorporating two stages dry fractionation when the PKL discount was greater than 2.43%, refinersand processors could boost their net margins. When the PKL discount was 9%, the net margin could beboosted by up to 16%.

. Introduction

The oil palm, scientifically known as Elaeis guineensis, is a veryersatile and unique industrial crop as it produces two distinctypes of oils from its fruits: palm-based oil from the mesocarp andalm kernel-based oil from the kernel (Malaysian Palm Oil Council,008, 2009). Palm kernel oil (PKO) when fractionated, yields palmernel olein (PKL) and palm kernel stearin (PKS). PKL finds its pri-ary use in non-food applications, especially as precursors to fatty

lcohols, fatty amines, fatty amides, glycerol and biodiesels. PKSnds its primary use in food applications, especially as precursor to

auric cocoa butter substitutes (CBS) (Choo et al., 2006; Malaysianalm Oil Council, 2009).

Traditionally, the PKO industry (refiners/processors) either dryr, to a less common extend, solvent fractionates PKO for stearin

raction with iodine value below seven, PKS IV7. They then hydro-enate it to produce the precursor to CBS. CBS has chemicallynd physically different glycerides from cocoa butter yet exhibits

∗ Corresponding author at: Department of Chemical and Environmental Engineer-ng, Faculty of Engineering, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor.E., Malaysia. Tel.: +60 3 8946 6288; fax: +60 3 8656 7120.

E-mail address: chuah@eng.upm.edu.my (T.-G. Chuah).

926-6690/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.indcrop.2013.05.019

© 2013 Elsevier B.V. All rights reserved.

comparable organoleptic properties. To an even lesser commonextend, the precursor to CBS could be produced through two alter-native routes: solvent fractionate PKO for stearin fraction withiodine value below five, PKS IV5; inter-esterify blends of PKO andcoconut oil along with lesser amount of non-lauric oils and thenhydrogenate them (Young, 1983; John Pease, 1985; Rossell, 1985;Traitler and Dieffenbacher, 1985; Md Ali and Dimick, 1994). Therehave been new developments in the use of supercritical technologyto fractionate PKO (Nik Norulaini et al., 2004; Norulaini et al., 2004;Zaidul et al., 2006). However, the use of supercritical technology inPKO is relatively rare when compared to uses in other vegetableoils (Akanda et al., 2012).

With the introduction and advancement of membrane filterpress technology, dry fractionation of edible oils can be carriedout with improved phase separation (Hamm, 1995). Static dryfractionation technology was developed and two stages static dryfractionation of PKO was used to produce PKS from which a majorpart could be used as non-hydrogenated precursor to CBS (Calliauwet al., 2005). This eliminated the concerns of trans-fat that resultedfrom the hydrogenation process. However, as PKO and its cor-

responding derivatives were soft commodities with varying andfluctuating price points, it was not always commercially feasible torun the double fractionation process. When the premiums for PKSor the discounts for PKL were not adequate to compensate each

438 G.F.L. Koay et al. / Industrial Crops and Products 49 (2013) 437– 444

ude P

oa

ot2etBestTpebe

2

2

pCbimtst

Fig. 1. Schematic diagram of cr

ther, running the double fractionation process would prove to be losing proposition for refiners and processors.

Though dry fractionation of PKO is an established technol-gy, past researches mainly assessed the economics in operatinghe hardware or running the process (Kellens, 2000; Calliauw,011; Kellens et al., 2012). This research, with the concepts andquations presented, addresses the dynamic and intrinsic rela-ionships between the three possible products of fractionation.y utilizing actual products prices, this work injects economicmphasis to the aforementioned dynamic and intrinsic relation-hips. These relationships play an important role in determininghe optimum operability opportunity of the fractionation process.his research would supplement the database for industrial scalelant production managements and process switchover consid-rations. The feasibility study would strengthen the theoreticalase especially when considering process modifications and plantxpansions.

. Methodology

.1. Dry fractionation of crude PKO

The fractionation of crude PKO was enacted in a dry fractionationilot unit (Fig. 1), similar to that reported by Calliauw et al. (2005).rude PKO was heated to 45 ◦C and homogenized in feed tank (A)efore being fed to dynamic crystallizer (B). It was pre-crystallized

n the dynamic crystallizer at 18–22 ◦C for 4 h, initiating the for-

ation of crystal nuclei. The crystal nuclei were left to grow and

o form cakes in static crystallizer (C) for 6 h. The crystallized PKOlurry was sent to a 30 bar membrane filter press (D) for dry frac-ionation. The liquid fraction, crude PKL, was transferred to product

KO dry fractionation pilot unit.

tank (E). The solid fraction, crude PKS, was melted at 60 ◦C in hop-per (F) and transferred to product tank (G). For the second stage ofthe two stages dry fractionation, crude PKL was re-crystallized at19 ◦C in the dynamic crystallizer (B) and the subsequent steps weresimilar to that of the first stage. The dry fractionation procedureswere carried out in quadruplicates and the averaged fractionationyield and cost were taken as the final values.

Three different pathways of crude PKO dry fractionation werecarried out: (1) dry fractionation of crude PKO for PKS IV7 and cor-responding PKL (one stage; dynamic crystallization: 18 ◦C, 4 h); (2)dry fractionation of crude PKO for PKS IV5 and corresponding PKL(one stage; dynamic crystallization: 22 ◦C, 4 h); (3) dry fractionationof crude PKO for PKS IV5 followed by dry fractionation of corre-sponding PKL for PKS IV7 (two stages; first dynamic crystallization:22 ◦C, 4 h; second dynamic crystallization: 19 ◦C, 4 h) (Fig. 2).

2.2. Deciphering price statistics

The monthly prices (in MYR/mt) of palm oil products tradedfrom January 2008 to December 2012 were obtained fromEconomics and Industry Development Division (2013). Malaysiabeing the international hub for palm trading, the primary and sec-ondary transaction currencies were in MYR and USD respectively.Since the transaction records for majority of crude grade palm oilproducts were in MYR while those of processed grades were in MYRand USD, MYR was used as the currency of choice for ease of cal-culation purpose. The calculated results could be easily converted

into USD with the prevailing average exchange rate of MYR3.15to USD1.00 for the abovementioned period. As for the results pre-sented in percentage, they would not be affected as percentage isa dimensionless number.

G.F.L. Koay et al. / Industrial Crops and Products 49 (2013) 437– 444 439

KO dr

fPmsspps

a

R

D

Fig. 2. One and two stages crude P

The selected palm oil products were: locally delivered crude PKOor Peninsular Malaysia and free on board basis refined grade PKO,KL and PKS IV7. The average prices of each product were deter-ined by arithmetic mean, assuming that by including trades from

pot to all forward months (spot +1 to spot +3), it was adequate andignificant representation of average prices, as the weighted meanrices were not available. The movement of the average monthlyrices of the four aforementioned feedstock and products is pre-ented in Fig. 3.

The price relationships between the four feedstock and productsre represented by Eqs. (1)–(3):

ef = RPKO(1)

1.05 − CPKO

isc = 1 − (RPKL − Ref)CPKO

× 100% (2)

Fig. 3. Movement of average (spot and forward) monthly prices rele

y fractionation process pathways.

Prem7 = RPKS7 − Ref − CPKO (3)

where CPKO, average price of crude PKO, MYR; RPKO, average priceof refined grade PKO, MYR; RPKL, average price of refined gradePKL, MYR; RPKS7, average price of refined grade PKS IV7, MYR; Ref,refining premium (regardless of product, inclusive of actual refiningcost plus markups), MYR; Disc, discounted value of PKL, %; Prem7,premium value of PKS IV7, MYR.

2.3. Feasibility assessment

The economic feasibility was assessed through a three waycounter comparison of premiums/discounts, returns/breakeven

and gains/losses of one and two stages dry fractionation of crudePKO. Under normal circumstances, PKS commands a premium(absolute value) over corresponding PKO to compensate for thediscount (in percentage) that is incurred by the corresponding PKL.

ased for palm oil products traded between Jan’08 and Dec’12.

440 G.F.L. Koay et al. / Industrial Crops and Products 49 (2013) 437– 444

nt on

Tstbyp

bofs

C

P

P

P

P

wooic

C

P

P

wooic

P

Fig. 4. Process flow of the feasibility assessme

his was counter tallied and verified with publicly available pricetatistics (60 months of palm oil products prices, traded from 2008o 2012) released by MPOB, the Malaysian government’s statutoryody overlooking palm oil industry. The actual dry fractionationield and cost were then amalgamated with the price statistics forrocess pathway feasibility assessment (Fig. 4)

Based on the fractionation yield and cost, the relationshipsetween the premiums required on PKS (PKS IV5 or PKS IV7) inrder to breakeven the discounts incurred by PKL in one stage dryractionation were established, are represented by Eqs. (4) and (5),implified and generalized into Eq. (6):

PKO + F1,S7 = Y1,S7 × PKS7 + Y1,L7 × PKL7 (4)

rice for PKS IV7,

KS7 = CPKO + P1,S7

rice for corresponding PKL,

KL7 = CPKO × (1 − D1,L7)%

here, CPKO, average price of crude PKO; F1,S7, fractionation costf one stage dry fractionation for PKS IV7; Y1,S7, fractionation yieldf single stage PKS IV7; Y1,L7, fractionation yield of correspond-ng PKL; P1,S7, breakeven premium of PKS IV7; D1,L7, discount oforresponding PKL.

PKO + F1,S5 = Y1,S5 × PKS5 + Y1,L5 × PKL5 (5)

Price for PKS IV5,

KS5 = CPKO + P1,S5

Price for corresponding PKL,

KL5 = CPKO × (1 − D1,L5)%

here, CPKO, average price of crude PKO; F1,S5, fractionation costf one stage dry fractionation for PKS IV5; Y1,S5, fractionation yieldf single stage PKS IV5; Y1,L5, fractionation yield of correspond-ng PKL; P1,S5, breakeven premium of PKS IV5; D1,L5, discount of

orresponding PKL.

Eq. (4) and Eq. (5) can be simplified and generalized to:

= [F + (1 − Y) × D × CPKO] ÷ Y (6)

one and two stages dry fractionation of PKO.

where, P, breakeven premium; F, fractionation cost; Y, fractionationyield of PKS; D, discount for PKL.

The relationships between the premiums required on PKS (PKSIV5 and PKS IV7) in order to breakeven the discounts incurredby PKL in two stages dry fractionation is represented by Eq. (7),expanded, rearranged and simplified into Eq. (8):

CPKO + F1,S5 + Y1,L5 × F2,S7

= Y1,S5 × PKS5 + Y2,S7 × PKS7 + Y2,L7 × PKL7 (7)

Price for PKS IV5,

PKS5 = CPKO + P1,S5

Price for PKS IV7,

PKS7 = CPKO + P2,S7

Price for corresponding PKL,

PKL7 = CPKO × (1 − D2,L7)%

where, CPKO, average price of crude PKO; F1,S5, fractionation costof first stage dry fractionation for PKS IV5; F2,S7, fractionation costof second stage dry fractionation for PKS IV7; Y1,S5, fractionationyield of first stage PKS IV5; Y1,L5, fractionation yield of correspond-ing first stage PKL; Y2,S7, fractionation yield of second stage PKSIV7; Y2,L7, fractionation yield of corresponding second stage PKL;P1,S5, breakeven premium of first stage PKS IV5; P2,S7, breakevenpremium of second stage PKS IV7; D2,L7, discount of correspondingsecond stage PKL.

Eq. (7) can be expanded, rearranged and simplified to:

F1 + F2 × (1 − Y1) = P1 × Y1 + P2 × Y2 × (1 − Y1) − D × CPKO

× (1 − Y1) × (1 − Y2) (8)

where, D, discount for PKL; F1, first stage fractionation cost;F2, second stage fractionation cost; P1, breakeven premium

for first stage PKS; P2, breakeven premium for second stagePKS; Y1, first stage PKS yield, thus (1 − Y1) the first stage PKL yield;Y2, second stage PKS yield, thus (1 − Y2) the second stage PKLyield.

ps and Products 49 (2013) 437– 444 441

3

3

pItmasnumteiiab

cwmvfdiTfprsatfp

taiobta

3

aoy(ffOIq

tsaoads

lve-

, tw

enty

-fou

r-

and

sixt

y-m

onth

aver

age

of

cru

de

PKO

, refi

nin

g

pre

miu

ms,

PKL

dis

cou

nts

and

PKS

IV7

pre

miu

ms,

and

the

corr

esp

ond

ing

brea

keve

n

pre

miu

ms

of

PKS

IV7

and

PKS

IV5.

6

mon

ths

Jan

’08

Ap

r’08

Jul’0

8

Oct

’08

Jan

’09

Ap

r’09

Jul’0

9

Oct

’09

Jan

’10

Ap

r’10

Jul’1

0

Oct

’10

Jan

’11

Ap

r’11

Jul’1

1

Oct

’11

Jan

’12

Ap

r’12

Jun

’08

Sep

’08

Dec

’08

Mar

’09

Jun

’09

Sep

’09

Dec

’09

Mar

’10

Jun

’10

Sep

’10

Dec

’10

Mar

’11

Jun

’11

Sep

’11

Dec

’11

Mar

’12

Jun

’12

Sep

’12

4164

.33

3782

.75

2616

.38

1885

.50

2244

.71

2475

.46

2469

.77

2782

.46

3135

.65

3413

.10

4263

.96

5618

.63

5766

.29

4556

.63

3778

.25

3888

.08

3913

.98

3385

.73

163.

6221

0.83

233.

8317

4.21

172.

89

125.

74

95.6

2

119.

36

104.

13

141.

31

156.

72

154.

99

220.

27

176.

73

124.

27

110.

41

181.

95

247.

58−1

.92

1.27

1.92

3.70

−2.4

8−0

.89

2.61

6.28

5.22

4.45

3.53

1.69

−0.0

5

−2.5

4

1.20

4.48

7.76

9.25

902.

5874

3.09

792.

8280

9.67

786.

01

810.

51

727.

36

849.

04

1037

.94

863.

22

1074

.85

791.

28

787.

55

837.

10

949.

65

1038

.15

1048

.66

1072

.82

−2.0

621

9.76

223.

1725

7.14

39.7

998

.25

248.

2143

9.05

419.

6639

9.53

397.

1030

1.05

130.

98−6

4.38

214.

57

438.

31

662.

53

678.

9524

.02

426.

21

432.

38

493.

98

99.8

9

205.

88

477.

78

823.

80

788.

65

752.

16

747.

74

573.

59

265.

23

−88.

98

416.

79

822.

47

1229

.01

1258

.78

, m7

−

P 1,S

7

904.

6452

3.33

569.

6655

2.53

746.

2271

2.26

479.

1540

9.99

618.

2846

3.68

677.

7649

0.24

656.

5890

1.47

735.

08

599.

84

386.

12

393.

87

6

mon

ths

12

mon

ths

24

mon

ths

60

mon

ths

Jul’1

2

Jan

’08

Jul’0

8

Jan

’09

Jul’0

9

Jan

’10

Jul’1

0

Jan

’11

Jul’1

1

Jan

’12

Jan

’08

Jan

’09

Jan

’10

Jan

’11

Jan

’08

Dec

’12

Dec

’08

Jun

’09

Dec

’09

Jun

’10

Dec

’10

Jun

’11

Dec

’11

Jun

’12

Dec

’12

Dec

’09

Dec

’10

Dec

’11

Dec

’12

Dec

’12

2640

.83

3390

.35

2430

.54

2357

.24

2802

.71

3699

.80

5015

.13

4772

.27

3846

.11

3277

.41

2822

.61

3077

.84

4236

.04

4024

.84

3499

.41

105.

14

198.

72

218.

45

137.

77

92.0

3

169.

79

171.

84

150.

10

153.

11

172.

89

193.

11

177.

48

171.

39

169.

41

191.

233.

784.

300.

06

0.34

3.79

5.31

1.99

1.09

4.54

7.85

3.95

4.16

3.55

5.03

4.93

1171

.25

847.

70

755.

49

751.

96

903.

80

983.

20

947.

86

789.

62

1002

.55

1080

.61

731.

86

832.

88

886.

58

883.

99

872.

0930

9.44

389.

0613

8.75

150.

09

320.

35

476.

57

309.

65

226.

71

438.

55

581.

95

332.

85

354.

38

396.

75

487.

21

435.

2658

8.81

733.

1627

9.32

299.

8960

8.59

891.

8358

9.18

438.

8182

2.90

1082

.89

631.

24

670.

29

747.

11

911.

13

816.

93, m

7

−

P 1,S

7

861.

81

458.

65

616.

74

601.

87

583.

45

506.

64

638.

22

562.

90

564.

00

498.

66

399.

01

478.

50

489.

83

396.

78

436.

84

valu

e

for

Dis

c

den

oted

that

PKL

was

fetc

hin

g

a

pri

ce

pre

miu

m

to

the

corr

esp

ond

ing

cru

de

PKO

feed

stoc

k.

valu

e

for

P 1,S

7an

d

P 1,S

5oc

curr

ed

wh

en

the

pre

miu

m

of

PKL

was

able

to

offs

et

the

cru

de

PKO

feed

stoc

k

and

pro

cess

ing

cost

s.

G.F.L. Koay et al. / Industrial Cro

. Results and discussion

.1. Deciphering price statistics

The 60 months palm oil products prices (comprising of therice statistics for crude PKO, refined grade PKO, PKL and PKS

V7) were deciphered to obtain 19 sets of six-month, 9 sets ofwelve-month, 4 sets of twenty-four-month and 1 set of sixty-

onth average of the refining premium (Ref), PKL discount (Disc)nd PKS IV7 premium (Prem7) using Eqs. (1)–(3), and are pre-ented in Table 1. The products prices were not deciphered forarrower time gaps (e.g. one-, two- or three-month) average val-es as PKO products were soft commodities with dynamic priceovements that had wide fluctuating magnitudes. Furthermore,

hey were being traded not only on spot (a trade where its deliv-rable or expiration is on the same month as it is being concludeds a spot month trade) but also on forward months (a trade wherets deliverable or expiration is on the immediate following months it is being concluded is a forward month trade, a spot +1)asis.

From the exercise, the Ref which included the actual refiningost plus markups ranged from MYR92.03 to MYR247.58. The Discas as high as 9.25% but there were occasions too when PKL com-anded premium over crude PKO. Five out of the 33 sets average

alues showed that PKL commanded premium over crude PKO,rom 0.05 to 2.54%. The possible causes of this situation includedisparity in supply-demand balance and sudden, sharp and signif-

cant drop in crude PKO prices after the conclusion of PKL trades.he PKS IV7 always commanded premium over crude PKO, rangingrom MYR727.36 to MYR1171.25, buffered from the effects of dis-arity in supply-demand balance and sudden, sharp and significantise in crude PKO prices after the conclusion of PKS IV7 trades. Theum of the value of the fractions should always exceed the feedstocknd processing costs (Calliauw, 2011). This led to the conclusionhat on a product-to-product basis, the burden in ensuring that aractionation regime was commercially feasible and profitable waslaced on PKS IV7.

There was direct correlation between the Disc and Prem7 andhis was most obvious when referring to the 9 sets of twelve-monthverage values. A simple liner regression resulted in R2 of 0.7893n value. Under normal circumstances, PKS commands premiumver corresponding PKO to compensate for the discount incurredy corresponding PKL. The established numbers and verified rela-ionships satisfied the preliminary requirement of the feasibilityssessment flow (refer to Fig. 4).

.2. Crude PKO dry fractionation yield and cost

For the one stage dry fractionation of crude PKO, the fraction-tion yield was 36.60% of PKS IV7 with 63.40% of PKL or 24.15%f PKS IV5 with 75.85% of PKL. There was virtually no processield loss and the recovery was 100%, in agreement with Calliauw2011). For the two stages dry fractionation, the corresponding PKLrom the fractionation to obtain PKS IV5 was re-crystallized and re-ractionated. The yield was 13.85% of PKS IV7 with 86.15% of PKL.verall, the two stages dry fractionation produced 24.15% of PKS

V5 with 10.51% PKS IV7 and 65.34% PKL, effectively reducing theuantity of the discounted PKL.

The one stage dry fractionation for PKS IV7 cost MYR49.90 whilehe one stage dry fractionation for PKS IV5 cost MYR66.40. The twotages dry fractionation costs comprised of MYR66.40 (first stage)nd MYR50.49 (second stage) or a total of MYR116.89. The details

f the fractionation yield and cost for each respective pathwayre tabulated in Table 2. The normalized cost for the second stagery fractionation for PKS IV7 was slightly higher than the onetage dry fractionation for PKS IV7. This was because the second Ta

ble

1Th

e

six-

, tw

e

Ave

rage

Star

t

mon

thEn

d

mon

th

CPK

O, M

YR

Ref

, MY

RD

isc,

%1

Prem

7,

MY

RP 1

,S7

, MY

R2

P 1,S

5, M

YR

2

Net

mar

gin

MY

R

>

Pre

Ave

rage

Star

t

mon

thEn

d

mon

th

CPK

O, M

YR

Ref

, MY

R

Dis

c,

%1

Prem

7,

MY

RP 1

,S7

, MY

R2

P 1,S

5, M

YR

2

Net

mar

gin

MY

R

>

Pre

1N

egat

ive

2N

egat

ive

442 G.F.L. Koay et al. / Industrial Crops an

Tab

le

2Y

ield

s

and

cost

s

of

dif

fere

nt

cru

de

PKO

dry

frac

tion

atio

n

pat

hw

ays.

Feed

:

cru

de

PKO

On

e

stag

e

dry

frac

tion

atio

n

Two

stag

es

dry

frac

tion

atio

n

(2n

d

stag

e)Fi

rst

stag

e

sim

ilar

to

that

of

Path

way

(2)

Path

way

(1)

PKS

IV7

Cor

resp

ond

ing

PKL

Path

way

(2)

PKS

IV5

Cor

resp

ond

ing

PKL

Path

way

(3)

PKS

IV7

Cor

resp

ond

ing

PKL

Frac

tion

atio

n

yiel

d, %

36.6

0

63.4

024

.15

75.8

513

.85

86.1

5Fr

acti

onat

ion

cost

(nor

mal

ized

to

MY

R/m

t

pro

du

ct)

(1)

Raw

mat

eria

ls

cost

Irre

leva

nt

as

ther

e

was

no

add

itiv

es,

pro

cess

aid

s,

chem

ical

s

or

effl

uen

ts

add

edfo

r

the

dry

frac

tion

atio

nIr

rele

van

t

as

ther

e

was

no

pro

cess

yiel

dlo

sses

(2)

Uti

liti

es

cost

(i)

Elec

tric

ity

Usa

ge

84

KW

H

119

KW

H

85

KW

HU

nit

cost

MY

R

0.32

MY

R

0.32

MY

R

0.32

Cos

t

MY

R

26.8

8M

YR

38.0

8M

YR

27.2

0(i

i)

Stea

m

Usa

ge

168

kg

185

kg

170

kgU

nit

cost

MY

R

0.06

MY

R

0.06

MY

R

0.06

Cos

t

MY

R

10.0

8

MY

R

11.1

0

MY

R

10.2

0(3

)

Labo

ur

cost

(20%

)

MY

R

7.39

MY

R

9.84

MY

R

7.48

(4)

Var

iabl

e

over

hea

ds

(15%

)

MY

R

5.55

MY

R

7.38

MY

R

5.61

Gro

ss

frac

tion

atio

n

cost

MY

R

49.9

0

MY

R

66.4

0 M

YR

50.4

9

d Products 49 (2013) 437– 444

stage dynamic crystallization had to be carried out at a slightlyhigher temperature to ensure the formation of stable crystals thatcould withstand the filter press. Secondly, as there were lesser PKSmolecules in the PKL feed, thus the phase separation efficiencydropped.

It was interesting to note the dynamic and intrinsic relation-ships between the products from the dry fractionation process.Even though the dry fractionation costs were fixed parameters, forperiod Jan’11 to Jun’11 when PKL was fetching an average of 0.05%premium over corresponding crude PKO, for one stage dry frac-tionation there was still a need for breakeven premium for PKS IV7valued at MYR130.98 or for PKS IV5 valued at MYR265.23 in orderfor the dry fractionation process to be commercially feasible andprofitable.

However, for period Apr’11 to Sep’11 when PKL was fetchingan average of 2.54% premium over corresponding crude PKO, forone stage dry fractionation, the breakeven for PKS IV7 dipped andbecame a discount, valued at MYR64.38; the breakeven for PKS IV5dipped and became discount too, valued at MYR88.98. This denotedthat the premium gained by PKL during this period was signif-icant and able to offset the crude PKO feedstock and processingcosts.

3.3. Feasibility assessment

From the dry fractionation run using the dry fractionation pilotunit, the actual fractionation yields and costs of different pathwayswere known. These were coupled with deciphered data to give:Y1,S7; Y1,L7 36.60%; 63.40%Y1,S5; Y1,L5 24.15%; 75.85%Y2,S7; Y2,L7 13.85%; 86.15%F1,S7 MYR49.90F1,S5 MYR66.40F2,S7 MYR50.49CPKO independent variableD1,L7 semi-independent variableD1,L5 semi-independent variableD2,L7 semi-independent variableP1,S7 dependent variableP1,S5 dependent variableP2,S7 dependent variable

3.3.1. Case study for six-month scenario: Apr’12–Sep’12In one stage dry fractionation, with crude PKO priced at

MYR3385.73, refined PKO priced at MYR3814.98 and refinedPKL priced at MYR3320.07, applying Eqs. (1) and (2), D1,L7 was9.25%.

Applying Eq. (4) or the simplified Eq. (6), the breakeven pre-mium for PKS IV7, P1,S7 was MYR678.95. For this period, the earlierestablished PKS market premium, Prem7, was MYR1072.82 (referto Table 1), thus there was an average net margin of MYR393.87.Applying Eq. (5) or the simplified Eq. (6), the breakeven premiumfor PKS IV5, P1,S5 was MYR1258.78.

In two stages dry fractionation, with crude PKO priced atMYR3385.73, refined PKO priced at MYR3814.98 and refined PKLpriced at MYR3320.07, applying Eqs. (1) and (2), D2,L7 was 9.25%.However, there were two unknowns: breakeven premiums for firststage PKS IV5 and second stage PKS IV7, P1,S5 and P2,S7 respectively.

Assuming P1,S5 remained at MYR1258.78, applying Eq. (7) or thesimplified Eq. (8), the corresponding P2,S7 reduced from MYR678.95to MYR51.31, saving MYR627.65. At overall 10.51% PKS IV7 yield, asaving of MYR65.94/mt feed.

Assuming P2,S7 remained at MYR678.95, applying Eq. (7) orthe simplified Eq. (8), the corresponding P1,S5 reduced from

MYR1258.78 to MYR985.75, saving MYR273.03. At overall 24.15%PKS IV5 yield, a saving of MYR65.94/mt feed.

From the aforementioned two interdependent assumptions,regardless of the distribution of the premiums, when Disc (D1,L7 and

G.F.L.

Koay

et al.

/ Industrial

Crops and

Products 49 (2013) 437– 444

443

Table 3Extra savings and relative losses in running one and two stages dry fractionation of PKO.

Average 6 months

Start month Jan’08 Apr’08 Jul’08 Oct’08 Jan’09 Apr’09 Jul’09 Oct’09 Jan’10 Apr’10 Jul’10 Oct’10End month Jun’08 Sep’08 Dec’08 Mar’09 Jun’09 Sep’09 Dec’09 Mar’10 Jun’10 Sep’10 Dec’10 Mar’11

P1,S5 (if P2,S7 unchanged), MYR 218.25 468.24 472.07 510.37 265.40 331.29 500.30 715.38 693.53 670.85 668.10 559.85Extra savings/earnings, MYR3 −194.23 −42.03 −39.70 −16.39 −165.52 −125.41 −22.52 108.42 95.12 81.31 79.64 13.74Normalized savings, MYR/mt feed3 −46.91 −10.15 −9.59 −3.96 −39.97 −30.29 −5.44 26.18 22.97 19.64 19.23 3.32P2,S7 (if P1,S5 unchanged), MYR 444.44 316.39 314.43 294.81 420.29 386.54 299.97 189.80 200.99 212.61 214.02 269.46Extra savings/earnings, MYR3 −446.50 −96.62 −91.26 −37.67 −380.50 −288.29 −51.76 249.25 218.67 186.93 183.08 31.58Normalized savings, MYR/mt feed3 −46.91 −10.15 −9.59 −3.96 −39.97 −30.29 −5.44 26.18 22.97 19.64 19.23 3.32Extra margins, %> normalized savings ÷ (Prem7 − P1,S7) Loss Loss Loss Loss Loss Loss Loss 6.39 3.72 4.23 2.84 0.68

Average 6 months 24 months

Start month Jan’11 Apr’11 Jul’11 Oct’11 Jan’12 Apr’12 Jul’12 Jan’08 Jan’09 Jan’10 Jan’11End month Jun’11 Sep’11 Dec’11 Mar’12 Jun’12 Sep’12 Dec’12 Dec’09 Dec’10 Dec’11 Dec’12

P1,S5 (if P2,S7 unchanged), MYR 368.18 148.01 462.39 714.55 967.25 985.75 569.31 590.22 625.51 667.71 769.66Extra savings/earnings, MYR3 −102.95 −236.99 −45.59 107.92 261.76 273.03 19.50 41.02 44.79 79.40 141.47Normalized savings, MYR/mt feed3 −24.86 −57.23 −11.01 26.06 63.22 65.94 4.71 9.91 10.82 19.18 34.16P2,S7 (if P1,S5 unchanged), MYR 367.64 480.42 319.39 190.22 60.78 51.31 264.62 238.54 251.43 214.22 162.00Extra savings/earnings, MYR3 −236.67 −544.80 −104.81 248.09 601.75 627.65 44.82 94.31 102.96 182.53 325.21Normalized savings, MYR/mt feed3 −24.86 −57.23 −11.01 26.06 63.22 65.94 4.71 9.91 10.82 19.18 34.16Extra margins, %> normalized savings ÷ (Prem7 − P1,S7) Loss Loss Loss 4.34 16.37 16.74 0.55 2.48 2.26 3.91 8.61

Average 12 months 60 months

Start month Jan’08 Jul’08 Jan’09 Jul’09 Jan’10 Jul’10 Jan’11 Jul’11 Jan’12 Jan’08End month Dec’08 Jun’09 Dec’09 Jun’10 Dec’10 Jun’11 Dec’11 Jun’12 Dec’12 Dec’12

P1,S5 (if P2,S7 unchanged), MYR 659.04 376.94 389.72 581.60 757.66 569.54 476.07 714.82 876.43 711.10Extra savings/earnings, MYR3 74.12 −97.62 −89.83 26.98 134.16 19.64 −37.26 108.08 206.47 105.82Normalized savings, MYR/mt feed3 17.90 −23.57 −21.69 6.52 32.40 4.74 −9.00 26.10 49.86 25.56P2,S7 (if P1,S5 unchanged), MYR 218.66 363.16 356.61 258.32 168.14 264.50 312.38 190.08 107.31 191.99Extra savings/earnings, MYR3 170.40 −224.41 −206.52 62.03 308.43 45.15 −85.67 248.47 474.64 243.27Normalized savings, MYR/mt feed3 17.90 −23.57 −21.69 6.52 32.40 4.74 −9.00 26.10 49.86 25.56Extra margins, %> normalized savings ÷ (Prem7 − P1,S7) 3.90 Loss Loss 1.12 6.40 0.74 Loss 4.63 10.00 5.85

3 Negative value for denoted that the two stages fractionation was incurring relative losses.

4 ops an

Dvgfa

iEtubRrmbg

faqtopcftfoatc

rtwfftwilvt2

pHncwmptmmo

44 G.F.L. Koay et al. / Industrial Cr

2,L7) was 9.25%, two stages dry fractionation of crude PKO to sal-age PKS IV7 was a commercially profitable process to undertake. Itenerated extra savings/earnings at the quantum of MYR65.94/mteed (Table 3), or an extra 16.74% margin on top of the MYR393.87verage net margin.

The possible extra savings/earnings (or relative losses incurred)n running the two stages dry fractionation is tabulated in Table 3.ven though the two stages dry fractionation of crude PKO effec-ively reduced the quantity of the discounted PKL by salvagingp to 13.85% of the softer stearin fraction, PKS IV7, out of it,ut the process was not always commercially feasible or logical.eferring to the six-month scenario, on hindsight, continuouslyunning the two stages dry fractionation process in year 2008,ost part of year 2009 and most part of year 2011 could not

ring in extra revenue. It was actually corroding the exiting mar-in.

The extra savings/earnings at the quantum of MYR3.32/mteed for the six-month period from Oct’10 to Mar’11, which waschieved through additional premium products, was only ade-uate to match the extra processing costs. It was not enougho compensate for the opportunity loss incurred. In order toperate the two stages dry fractionation process, the through-ut of the fractionation unit was effectively halved. The originalrystallization cycle time of 10 h (4 h of dynamic crystallizationollowed by 6 h of static crystallization) had to be doubled. Fur-hermore, this has not taken into consideration the down timeor process switchover and the extra time needed for the sec-nd stage fractionation over at the membrane filter press. Therelso arose the need for at least an extra storage tank for an extraype of end product, thus contributing to higher fixed overheadosts.

However, the market situation changed and the prospects ofunning the two stages dry fractionation process from Oct’11o Sep’12 period were promising. This turn of event coincidedith the widening discount incurred by PKL. The two stages dry

ractionation process was actually salvaging premium PKS IV7rom the discounted PKL. When PKL was heavily discounted inhe market, it was good opportunity to run the process. Thereas strong direct correlation between Disc and the extra sav-

ngs/earnings by running the two stages dry fractionation. Theinear regression between these two resulted in R2 of 0.9451 inalue. Based on the current available data, the threshold in runninghe two stages dry fractionation was when Disc was greater than.43%.

It appears that continuously running the double fractionationrocess for 60 months would guarantee extra margin of up to 5.85%.owever, operating the fractionation process in such manner didot incorporate optimization, which is a core consideration in pro-ess engineering. As per highlighted earlier, there were stretchesithin the 60-month period (year 2008, most of year 2009 andost of year 2009) that actually introduced loss and corroded

rofit margins. Since PKO and its derivative are soft commodi-

ies that could be traded in forward months’ basis, the forward

onths’ prices should be considered as input when utilizing theathematical model to determine possible optimum operability

pportunities.

d Products 49 (2013) 437– 444

4. Conclusion

One stage dry fractionation of crude PKO was carried out andthe fractionation yield was 36.60% of PKS IV7 with 63.40% of PKL or24.15% of PKS IV5 with 75.85% of PKL. The two stages dry fraction-ation yielded 24.15% of PKS IV5 and 10.51% PKS IV7 with the restbeing PKL, in a single cycle. There was direct correlation betweenthe PKL discount and PKS premium. When PKL was commandingpremium over crude PKO, it was better to single fractionate crudePKO. When PKL was heavily discounted, it was good opportunity toproceed with two stages fractionation. Running the two stages frac-tionation process when there was wide discount for PKL providedextra margin of up to 16.74%. Based on this study, the threshold inrunning two stages dry fractionation was when the PKL discountwas greater than 2.43%. Lastly, the concepts and feasibility studypresented in this research shall form a strong theoretical base inassisting future process (re)evaluation and decision making espe-cially in process switchover considerations and plant expansionplanning amidst dynamic market movements.

References

Akanda, M.J.H., Sarker, M.Z.I., Ferdosh, S., Manap, M.Y.A., Rahman, N.N.N.A., Kadir,M.O.A., 2012. Applications of supercritical fluid extraction (SFE) of palm oil andoil from natural resources. Molecules 17, 1764–1794.

Calliauw, G., Foubert, I., De Greyt, W., Dijckmans, P., Kellens, M., Dewettinck, K., 2005.Production of cocoa butter substitutes via two-stage static fractionation of palmkernel oil. J. Am. Oil Chem. Soc. 82, 783–789.

Calliauw, G., Dry fractionation. In: Dijkstra, A.J., (Ed.) The AOCS Lipid Library,American Oil Chemists’ Society. [Online] available at: http://lipidlibrary.aocs.org/processing/dryfract/index.htm (accessed 31.10.12).

Choo, Y.M., Ma, A.N., Basiron, Y., Yung, C.L., Cheng, S.F., 2006. Palm-based biodieselfoundation. US Patent 0,288,637 (2006).

Economics and Industry Development Division, Malaysian Palm Oil Board. MPOBmonthly prices of palm oil products traded from 2008 to 2012. [Online]available at: http://bepi.mpob.gov.my/index.php/statistics/price/monthly.html(accessed 29.04.13).

Hamm, W., 1995. Trends in edible oil fractionation. Trends Food Sci. Technol. 6,121–126.

John Pease, J., 1985. Confectionery fats from palm oil and lauric oil. J. Am. Oil Chem.Soc. 62, 426–430.

Kellens, M., 2000. Oil modification processes. In: Hamm, W., Hamilton, R.J. (Eds.),Edible Oil Processing. Sheffield Academic Press, Sheffield, pp. 129–173.

Kellens, M., Hendrix, M., Calliauw, G., 2012. Crystallization apparatus and processfor molten fats. US Patent 8,133,519 (2012).

Malaysian Palm Oil Council, 2008. Palm Oil & Palm Kernel Oil Applications. MalaysianPalm Oil Council, Malaysia.

Malaysian Palm Oil Council, Malaysian Palm Oil Board, 2009. Fact Sheets: MalaysianPalm Oil. Malaysian Palm Oil Council and Malaysian Palm Oil Board, Malaysia.

Md Ali, A.R., Dimick, P.S., 1994. Melting and solidification characteristics of confec-tionery fats: anhydrous milk fat, cocoa butter and palm kernel stearin blends. J.Am. Oil Chem. Soc. 71, 803–806.

Nik Norulaini, N.A., Md Zaidul, I.S., Anuar, O., Mohd Omar, A.K., 2004. Supercriticalenhancement for separation of lauric acid and oleic acid in palm kernel oil (PKO).Sep. Purif. Technol. 39, 133–138.

Norulaini, N.A.N., Zaidul, I.S.M., Anuar, O., Azlan, A., Omar, A.K.M., 2004. Supercrit-ical reduction of lauric acid in palm kernel oil (PKO) to produce cocoa butterequivalent (CBE). J. Chem. Eng. Jpn. 37, 194–203.

Rossell, J.B., 1985. Fractionation of lauric oils. J. Am. Oil Chem. Soc. 62, 385–390.Traitler, H., Dieffenbacher, A., 1985. Palm oil and palm kernel oil in food products. J.

Am. Oil Chem. Soc. 62, 417–421.

Young, F.V.K., 1983. Palm kernel and coconut oils: analytical characteristics, process

technology and uses. J. Am. Oil Chem. Soc. 60, 374–379.Zaidul, I.S.M., Nik Norulaini, N.A., Mohd Omar, A.K., Smith Jr., R.L., 2006. Supercritical

carbon dioxide (SC-CO2) extraction and fractionation of palm kernel oil frompalm kernel as cocoa butter replacers blend. J. Food Eng. 73, 210–216.