BagajewiczHeatIntegration PDF

PART 1
PINCH AND MINIMUM

UTILITY USAGE
WHAT IS THE PINCH?
• The pinch point is a temperature.
• Typically, it divides the temperature range
into two regions.
• Heating utility can be used only above the
pinch and cooling utility only below it.
WHAT IS A PINCH DESIGN
• A heat exchanger network obtained using
the pinch design method is a network where
no heat is transferred from a hot stream
whose temperature is above the pinch to a
cold stream whose temperature is below the
pinch.
IS PINCH TECHNOLGY
CURRENT?
• YES and NO.
• It is a good first approach to most problems.
• Pinch technology is at the root of many other
heat integration technologies. It is impossible
to understand them without the basic concepts
of pinch technology.
TEMPERATURE-ENTHALPY
(T-H) DIAGRAMS
• Assume one heat exchanger. These are
T
alternative representations
H,in
TC,in T TH,in
TC,out
Q
TC,out
TH, out
TH, out Slopes are the
TH,in TH, out TC,in inverse of F*Cp.
Recall that Q=F Cp ∆T
TC,out TC,in ∆H
Q
Q
T-H DIAGRAMS
• Assume one heat exchanger and a heater
TH,in
TC,in TC,out T TH,in

Q QH TC,out
TH, out
TH, out
TH,in TH, out TC,in
TC,out ∆H
TC,in
H Q QH
QH Q
T-H DIAGRAMS
• Assume one heat exchanger and a cooler
TH,in
TC,in TC,out
Q
T TH,in
TC,out
QC
TH, out TH, out
TH,in TH, out TC,in
C
QC
TC,out ∆H
TC,in QC Q
Q
T-H DIAGRAMS
• Two hot-one cold stream
TH1,in
TH2,in
T TH2,in
TC,in TC,out
Q1 Q2 TH1,in
TC,out
TH2,out TH2,out
TH2,out
TH1,in TH1,out TH1,out

TH2,in TH2, out TC,in
TC,out ∆H
TC,in
Q1 Q2
Q2 Q1
Notice the vertical arrangement of heat transfer
T-H DIAGRAMS
• Composite Curve
Obtained by lumping all the heat from different streams that are
at the same interval of temperature.
T T
∆H ∆H
Remark: By constructing the composite curve we loose information on
the vertical arrangement of heat transfer between streams
T-H DIAGRAMS
• Moving composite curves horizontally
T T
Cooling
Heating
∆H ∆H
Smallest ∆T Smallest ∆T
TH1,in TH2,in
TH1,in TH2,in TC,in
TC,in Q1 Q2 QH TC,out
Q1 Q2
TC,out TH2,out
QC
TH1,out TH2,out TH1,out
T-H DIAGRAMS
T Moving the cold composite
stream to the right
Cooling
• Increases heating and cooling BY
THE SAME AMOUNT
Heating • Increases the smallest ∆T
• Decreases the area needed
Smallest ∆H A=Q/(U* ∆T )
∆T TH1,in TH2,in
TC,in Notice that for this simple
Q1 Q2 QH TC,out
example the smallest ∆T
takes place in the end of the
QC TH2,out
cold stream
TH1,out
T-H DIAGRAMS
T • In general, the smallest ∆T

Cooling can take place anywhere.
Heating
• We call the temperature at
which this takes place THE
PINCH.
∆H
TEMPERATURE-ENTHALPY
DIAGRAMS
• From the energy point
T
of view it is then
convenient to move the
Cooling cold stream to the left.
• However, the area may
become too large.
Heating
• To limit the area, we
∆H introduce a minimum
approach ∆Tmin
GRAPHICAL PROCEDURE
• Fix ∆Tmin
• Construct the hot and cold composite curve
• Draw the hot composite curve and leave it fixed
• Draw the cold composite curve in such a way that
the smallest ∆T=∆Tmin
• The temperature at which ∆T=∆Tmin is the PINCH
• The non-overlap on the right is the Minimum
Heating Utility and the non-overlap on the left is the
Minimum Cooling Utility
HANDS ON EXERCISE
∆H=27 MW ∆H=-30 MW
REACTOR 2
T=140 0C T=230 0C T=80 0C
0
T=200 C
∆H=32 MW ∆H=-31.5 MW
REACTOR 1
T=20 0C T=40 0C
T=180 0C T=250 0C
Stream Type Supply T Target T ∆H F*Cp

(oC) (oC) (MW) (MW oC-1)
Reactor 1 feed Cold 20 180 32.0 0.2
Reactor 1 product Hot 250 40 -31.5 0.15
∆Tmin=10 oC
Answer: Hot Streams
250 250
200 200
15
5
.
0.2 FCp=0.15
p=0
.4
=0
p=
p
FC
FC
FC
80 80
40 40 FCp=0.15
31.5 30 ∆H 6 48 7.5 ∆H
Answer: Cold Streams
0 .3
230 .3 230 p=
= 0 F C
Cp
180 F 180 =0.5
F Cp
140 140
.2
0.2
=0
p=
p
FC
FC
20 20
∆H ∆H
32 27 24 20 15
Answer: Both Curves Together.
250
230
200
180 ∆T= ∆Tmin
Pinch
140
80
40
20
∆H
10 51.5 7.5
Important observation: The pinch is at the beginning of a cold stream or at
the beginning of a hot stream.
UTILITY COST vs. ∆Tmin
• There is total overlap for some values of ∆Tmin
T
COST
∆H
Utility
TOTAL OVERLAP
∆Tmin ∆H
PARTIAL OVERLAP
Note: There is a particular overlap that requires only cooling utility

PROBLEM TABLE
• Composite curves are inconvenient. Thus
a method based on tables was developed.
• STEPS:
1. Divide the temperature range into intervals and
shift the cold temperature scale
2. Make a heat balance in each interval
3. Cascade the heat surplus/deficit through the
intervals.
4. Add heat so that no deficit is cascaded
PROBLEM TABLE
• We now explain each step in detail.
Consider the example 1.1

∆Tmin=10 oC
PROBLEM TABLE
1. Divide the temperature range into intervals and shift the
cold temperature scale
250
250
240
230
200 200
190
180
150
140
80 80
40 40
30
20
Hot Cold Hot Cold
streams streams streams streams
Now one can make heat balances in each interval. Heat transfer within
each interval is feasible.
PROBLEM TABLE
2. Make a heat balance in each interval. (We now turn into
a table format distorting the scale)
∆Tinterval ∆Hinterval Surplus/Deficit?
250 F Cp=0.15
10 1.5 Surplus
240
F Cp=0.25 40 - 6.0 Deficit
200
10 1.0 Surplus
190
40 -4.0 Deficit
150
F Cp=0.3 70 14.0 Surplus
80
40 -2.0 Deficit
40
30 10 - 2.0 Deficit
F Cp=0.2
Hot Cold
streams streams
PROBLEM TABLE
3. Cascade the heat surplus through the intervals. That is,
we transfer to the intervals below every surplus/deficit.
1.5 1.5
This interval has a 1.5 The largest deficit
- 6.0 surplus. It should - 6.0 transferred is -7.5.
transfer 1.5 to
interval 2. -4.5 Thus, 7.5 MW of
1.0 1.0 heat need to be
This interval has a -3.5 added on top to
-4.0 deficit. After using prevent any deficit
-4.0
the 1.5 cascaded it to be transferred to
-7.5 lower intervals
transfers –4.5 to
14.0 14.0
interval 3.
6.5
- 2.0 -2.0
4.5
-2.0 - 2.0
2.5
PROBLEM TABLE
4. Add heat so that no deficit is cascaded.
7.5
1.5 1.5 This is the

1.5 9.0 minimum heating
- 6.0 - 6.0 utility
-4.5 3.0
1.0 1.0 This is the

-3.5 4.0 position of the
-4.0 -4.0 pinch
-7.5 0.0
14.0 14.0
14.0
This is the
6.5
minimum cooling
-2.0 -2.0
utility
4.5 12.0
-2.0 -2.0
2.5 10.0
IMPORTANT CONCLUSION
7.5 + λ
DO NOT TRANSFER 1.5 Heating utility is

HEAT ACROSS THE 9.0 + λ larger than the
minimum
- 6.0
PINCH 3. 0 + λ
THIS IS A GOLDEN RULE OF PINCH 1.0 Heat is
TECHNOLOGY. 4. 0 + λ transferred
•WE WILL SEE LATER HOW THIS IS -4.0 across the pinch
RELAXED FOR DIFFERENT 0. 0 + λ
PURPOSES
14.0
• WHEN THIS HAPPENS IN BADLY Cooling utility is
14. 0 + λ
INTEGRATED PLANTS THERE ARE larger by the same
HEAT EXCHANGERS WHERE SUCH
-2.0 amount
TRANSFER ACROSS THE PINCH 12. 0 + λ
TAKES PLACE -2.0
10. 0 + λ
PROBLEM TABLE
Heating utility of smaller temperature. Heating utility at
0.0 the largest
7.5
temperature is
1.5
1.5 now zero.
1.5 + 4.5
9.0
- 6.0
- 6.0
0.0
3.0
1.0
1.0 These are the
4.0
1.0 + 3.0 minimum values
-4.0 of heating utility
-4.0
0.0
needed at each
0.0
14.0
temperature
14.0
level.
14.0
14.0
-2.0
-2.0
12.0
12.0
-2.0
-2.0
10.0
10.0
MATHEMATICAL MODEL
δ0 Let q i be the surplus or demand of heat in interval i .
It is given by:
q1
δ1
qi = ∑F k
H
cp kH (Ti −1 − Ti ) − ∑F s
C
cp sC (Ti −1 − Ti )
k ∈ΓiH s∈ΓiC
q2
The minimum heating utility is obtained by solving
the following linear programming (LP) problem
qi S min = Min δ 0
δi
qi+1 s. t
δ i = δ i −1 + q i ∀i = 1,... mI
δi+1
δi ≥0
qn
δn
PART 2
TOTAL AREA TARGETING

In this part we will explore ways to predict the total area of a network
without the need to explore specific designs.
Because A=Q/(U*∆Tml), one can calculate the area easily in the
following situation.
T
TH1
TC2
TH2 (TH1-TC2)-(TH2-TC1)
∆Tml=
(TH1-TC2)
ln
(TH2-TC1)
TC1
∆H
Q
Since area=Q/(U ∆Tml), the composite curve diagram provides one
way of estimating the total area involved. Isolate all regions with a
pair of straight line sections and calculate the area for each.
T
Heating utility,
steam or furnace.
Cooling
water
∆H
A1 A2 A3 A4 A5 A6
The above scheme of heat transfer is called VERTICAL HEAT TRANSFER

EXERCISE
Calculate the values of Q in each interval and estimate the
corresponding area. Use U= 0.001 MW m-2 oC
T Furnace (300 oC)
250
230
200
180 ∆T= ∆Tmin
Pinch
140
80
Obtain the total
40 area estimate
20
Cooling
Water ∆H
Q1 Q2 Q3 Q4 Q5 Q6
EXERCISE
COMPOSITE CURVE Furnace
(300 oC)
300
250
200 I II III IV
T, C
Pinch 150 V VI
100
Cooling 50 ∆T= ∆Tmin

Water
0
0 10 20 30 40 50 60 70 80 90
Q, MW
HOT COLD
Units:
Interval Q TH1 TH2 TC1 TC2
Q= MW
I 6 80 40 15 20
T= oC
II 4 90 80 20 30
A= m2 III 24 150 90 20 140
IV 20 200 150 140 180
V 7.5 250 200 180 205
VI 7.5 300 250 205 230
EXERCISE
Interval Q TH1 TH2 TC1 TC2 ∆Tml A

I 6 80 40 15 20 40.0 150.1
II 4 90 80 20 30 60.0 66.7
III 24 150 90 20 140 30.8 778.4
IV 20 200 150 140 180 14.4 1386.3
V 7.5 250 200 180 205 30.8 243.3
VI 7.5 300 250 205 230 81.9 91.6
Total Area 2716.3
Units: Q= MW T= oC , A= m2
U= 0.001 MW m-2 oC
Drawbacks
• Fixed costs associated with the number of units
are not considered.
We will see later how the number of units can be

calculated
QUESTIONS FOR DISCUSSION
• Is the total area predicted this way, realistic?

That is, is it close enough to a value that one
would obtain in a final design?
• Is the estimate, realistic or not, conservative?
That is, is it larger than the one expected from a
final design?
•How complex is a design built using the vertical
transfer?
ANSWERS
• Is the total area predicted this way, realistic?
That is, is it close enough to a value that one
would obtain from a final design?
YES, Within 10-15%

ANSWERS
•Is the estimate, realistic or not, conservative?
That is, is it larger than the one expected from
a final design?
The area obtained is actually the minimum

area needed to perform the heat transfer.
ANSWERS
•How complex is a design built using the
vertical transfer?
Very Complex. Take for example interval 4.
There are four streams in this interval.
(oC) (oC) (MW)
(MW oC-1)

This implies at least three heat exchangers,

just in this interval.
HEAT EXCHANGER NETWORK
300
40 80 90 150 200 250 300
R1 prod, FCp=0.15
FCp=0.09
FCp=0.16
R2 prod, FCp=0.25
FCp=0.1875
R1 feed, FCp=0.2
R2 feed, FCp=0.3
FCp=0.1125
I II III IV V VI
15 20 30 140 180 205 230
20
HEAT EXCHANGER NETWORK
40 80 90 150 200 250 300
300
R1 prod, FCp=0.15
FCp=0.16
R2 prod, FCp=0.25
20 R1 feed, FCp=0.2
FCp=0.075
R2 feed, FCp=0.3
FCp=0.1125
I II III IV V VI
15 20 30 140 180 205 230
TOTAL= 10 Exchangers
Called “Spaghetti” design
PREDICTING THE NUMBER OF
UNITS
We can anticipate very simply how many exchangers we should have!!!
Consider the following warehouses, each containing some merchandise

that needs to be delivered to the row of consumer centers. What is the
minimum number of trucks needed?
Warehouses 30 50 17
Consumer 25 16 56
Centers
ANSWER
You need five trucks, possibly less in some other cases. Here is how
you solve the problem specifically.
30 50 17
25 5 11 39 17
25 16 56
The general answer is N=S-1 . When does one need less?

ANSWER
When there is an exact balance between two streams or a
subset of streams.
25 55 17
25 16 39 17
25 16 56
The general answer is N=S-P . P is the number of

independent subsystems. (Two in this case)
GENERAL FORMULA FOR
UNIT TARGETING
Nmin= (S-P)above pinch+ (S-P)below pinch
If we do not consider two separate problems, above

and below the pinch we can get misleading results.
SUPERTARGETING
• Economy of the system is dependent on ∆Tmin
COST
Total
Utility
Capital
Optimum ∆Tmin
SPECIAL CASES
• There is total overlap for some values of ∆Tmin
T
COST
Total ∆H
TOTAL
Utility OVERLAP
Capital
∆H
PARTIAL OVERLAP
∆Tmin
Note: There is a particular overlap that requires only cooling utility
PART 3
DESIGN OF MAXIMUM
ENERGY RECOVERY
NETWORKS
MER NETWORKS
• Networks featuring minimum utility usage
are called MAXIMUM ENERGY
RECOVERY (MER) Networks.
PINCH DESIGN METHOD
RECALL THAT
• No heat is transferred through the pinch.
• This makes the region above the pinch a
HEAT SINK region and the region below
the pinch a HEAT SOURCE region.
Heat Sink
Minimum
Heat is obtained from 7.5
heating utility
the heating utility 1.5
9.0
-6.0
T
3.0
1.0
4.0
-4.0
0.0 Pinch
14.0
14.0
-2.0
12.0
H
-2.0
Minimum
Heat Source 10.0
cooling utility
Heat is released to
cooling utility
CONCLUSION
• One can analyze the two systems separately,
that is,
• Heat exchangers will not contain heat

transfer across the pinch.
PINCH MATCHES
• Consider two streams above the pinch
TC,out= Tp- ∆Tmin + Q/FCpC
TH,in Tp
TH,in= Tp+Q/FCpH
But TH,in> TC,out+ ∆Tmin.
Thus replacing one obtains Q/FCpH > Q/FCpC
TC,out Q Tp- ∆Tmin FCpH < FCpC Golden rule for pinch
matches above the pinch.
∆Tmin ∆Tmin Violation when FCpH > FCpC

PINCH MATCHES
•Consider two streams below the pinch
TC,in= Tp- ∆Tmin - Q/FCpC
Tp TH,out
TH,out= Tp-Q/FCpH
But TH,out> TC,in + ∆Tmin.
Thus replacing one obtains
Q TC,in Golden rule for pinch

FCpC < FCpH
Tp- ∆Tmin matches below the pinch.
∆Tmin ∆Tmin
Violation when FCpC > FCpH

CONCLUSION
• Since matches at the pinch need to satisfy these rules,
one should start locating these matches first. Thus, our
first design rule:
START BY MAKING PINCH

MATCHES
QUESTION
• Once a match has been selected how

much heat should be exchanged?
ANSWER
• As much as possible!
• This means that one of the streams has its duty
satisfied!!
THIS IS CALLLED THE
TICK-OFF RULE
HANDS ON EXERCISE
∆H=27 MW ∆H=-30 MW
T=230 0C REACTOR 2
T=140 0C T=80 0C
T=200 0C
∆H=32 MW ∆H=-31.5 MW
REACTOR 1
T=20 0C
T=250 0C T=40 0C
T=180 0C

∆Tmin=10 oC PINCH=150 oC
HANDS ON EXERCISE
250 0C 150 0C 40 0C
FCp=0.15 H1
200 0C
80 0C
FCp=0.25 H2
180 0C 140 0C 20 0C
FCp=0.2 C1
FCp=0.3 230 0C C2

∆Tmin=10 oC PINCH=150 oC
ABOVE THE PINCH
250 0C 150 0C
FCp=0.15 H1
200 0C
FCp=0.25 H2
180 0C
C1
FCp=0.2
230 0C
FCp=0.3
C2
140 0C
• Which matches are possible?

ANSWER (above the pinch)
150 0C
250 0C
FCp=0.15 H1
200 0C
FCp=0.25 H2
180 0C
FCp=0.2 C1
230 0C
FCp=0.3 C2
140 0C
• The rule is that FCpH < FCpC . We therefore can only

make the match H1-C1 and H2-C2.
ANSWER (above the pinch)
250 0C 203.3 0C 150 0C
FCp=0.15 H1
200 0C
FCp=0.25 H2
180 0C
FCp=0.2 C1
230 0C 181.7 0C 8
FCp=0.3 C2
140 0C
12.5
• The tick-off rule says that a maximum of 8 MW is exchanged

in the match H1-C1 and as a result stream C1 reaches its target
temperature.
• Similarly 12.5 MW are exchanged in the other match and the
stream H2 reaches the pinch temperature.
BELOW THE PINCH
150 0C 40 0C
FCp=0.15 H1
FCp=0.25 H2 80 0C
FCp=0.2 140 0C 20 0C C1
• Which matches are possible?

The rule is that FCpC < FCpH . Thus, we can
only make the match H2-C1
ANSWER (below the pinch)
FCp=0.15 H1 150 0C 40 0C
FCp=0.25 H2 80 0C
FCp=0.2 140 0C 52.5 0C 20 0C

C1
17.5
• The tick-off rule says that a maximum of 17.5 MW is

exchanged in the match H2-C1 and as a result stream H2
reaches its target temperature.
COMPLETE NETWORK
AFTER PINCH MATCHES
250 0C 203.3 0C 150 0C 40 0C
FCp=0.15 H1
H2 200 0C 80 0C
FCp=0.25
140 0C
180 0C 52.5 0C
FCp=0.2 C1
20 0C
181.7 0C 8 17.5
FCp=0.3 230 0C C2
140 0C
12.5
• Streams with unfulfilled targets are colored.

WHAT TO DO NEXT?
203.3 0C 150 0C 40 0C
FCp=0.15 H1 250 0C
H2 200 0C 80 0C
FCp=0.25
140 0C
180 0C 52.5 0C
FCp=0.2 C1
20 0C
181.7 0C 8 17.5
FCp=0.3 230 0C
C2
140 0C
12.5
• Away from the pinch, there is more flexibility to make

matches, so the inequalities do not have to hold.
• The pinch design method leaves you now on your own!!!!!
• Therefore, use your judgment as of what matches to select!!
ANSWER
250 0C 203.3 0C 150 0C 40 0C
FCp=0.15 H1
H2 200 0C 80 0C
FCp=0.25
140 0C
180 0C 52.5 0C
FCp=0.2 C1
20 0C
181.7 0C 8 17.5
FCp=0.3 230 0C C2
140 0C
12.5
• We first note that we will use heating above the pinch. Thus
all hot streams need to reach their inlet temperature. We are
then forced to look for a match for H1. Please locate it.
ANSWER
• The match is H1-C1. We finally put a heater on the
cold stream
250 0C 150 0C 40 0C
FCp=0.15 H1
H2 200 0C 80 0C
FCp=0.25
140 0C
180 0C 52.5 0C
FCp=0.2 C1
20 0C
230 0C 8 17.5
FCp=0.3 C2
H
140 0C
7.5 7 12.5
ANSWER
• Below the pinch we try to have the cold streams start at
their inlet temperatures and we later locate coolers (one in
this case).
250 0C 150 0C 40 0C
FCp=0.15 H1 C
10
H2 200 0C
FCp=0.25
80 0C
180 0C 140 0C 20 0C
FCp=0.2 C1
230 0C 8 17.5 6.5

FCp=0.3 C2
H
140 0C
7.5 7 12.5
EXAMPLE 150 0C
FCp=0.15 H1 250 0C 40 0C
C
10
H2 200 0C
FCp=0.25
80 0C
180 0C 140 0C 20 0C
FCp=0.2 C1
230 0C 8 17.5 6.5

FCp=0.3 C2
H
140 0C
7.5 7 12.5
Nmin= (S-P)above pinch+ (S-P)below pinch =

=(5-1) + (4-1) = 7
If we do not consider two separate problems
Nmin= (6-1)= 5, which is wrong
Note: A heat exchanger network with 5 exchangers exists, but it is impractical and
costly. This is beyond the scope of this course.
UNEQUAL NUMBER OF
STREAMS AT THE PINCH
Indeed, if the number of hot streams is larger than the
number of cold streams, then no pinch matches are
possible.
150 OC 100 OC Assume the matches H1-C1 and
FCp=0.15 H1
the matches H2-C2 have been
FCp=0.25 140 OC
H2 selected. Since H3 needs to go to
130 OC the pinch temperature, there is no
FCp=0.1 H3
cold stream left to match, even if
127.5 OC 90 OC there is portions of C1 or C2 that
Target=170 OC FCp=0.2
C1 are left for matching. Such
122.5 OC
7.5 matching would be infeasible.
Target=140 OC FCp=0.4 C2
10
What is then, the solution?

ANSWER
Split cold stream until the inequality is satisfied.
150 OC 100 OC
FCp=0.15 H1
140 OC
FCp=0.25 H2
130 OC
FCp=0.1 H3
127.5 OC 90 OC
130 OC 7.5
FCp=0.15 110 OC 10
Notice that different combinations of flowrates in the split

satisfy the inequality.
INEQUALITY NOT
SATISFIED
Consider the following case:
150 OC 100 OC
FCp=0.5 H1
90 OC
ANSWER
Split the hot stream
FCp=0.2 H1 150 OC 100 OC
FCp=0.3
140 OC 90 OC
C1
127.5 OC 10
C2
15
SOLVE THE FOLLOWING
PROBLEM
Below the Pinch :
100 OC
FCp=0.5 H1 Target=40 OC
90 OC 30 OC
FCp=0.7 C1
ANSWER
Below the Pinch :
100 OC 40 OC
60 OC
90 OC 30 OC
FCp=0.5 C1
30
FCp=0.2
12
COMPLETE PROCEDURE
ABOVE THE PINCH
Start
Yes FCpH≤FCpC Yes

SH≤SC
at pinch?
No No
Split Cold
Stream
Place Split Hot

matches Stream
COMPLETE PROCEDURE
BELOW THE PINCH
Start
Yes FCpH ≥ FCpC Yes

SH≥SC
at pinch?
No No
Split Hot
Stream
Place
Split Cold
matches
Stream
HANDS ON EXERCISE
Type Supply T Target T F*Cp
(oC) (oC) (MW oC-1)
Hot 750 350 0.045
Hot 550 250 0.04
Cold 300 900 0.043
Cold 200 550 0.02
∆Tmin=50 oC
Minimum Heating Utility= 9.2 MW
Minimum Cooling Utility= 6.4 MW
ANSWER
FCp=0.04 550 0C
750 0C 358.89 0C
350 0C
FCp=0.045 H1
0.4
250 0C
FCp=0.04 H2
686.05 0C 6
400 0C
300 0C
FCp=0.043 900 0C C1
9.2 8 8.6
FCp=0.02 550 0C C2
200 0C
1 500 0C 6
TRANSSHIPMENT MODEL
(Papoulias and Grossmann, 1983)
δ0
We will now expand the mathematical model
q1 we presented to calculate the minimum utility.
δ1
q2 S min = Min δ 0
s. t
δ i = δ i −1 + q i ∀i = 1,... mI
qi δi ≥0
δi Where
qi+1
δi+1 qi = ∑ k k (Ti −1 − Ti ) −
F H
cp H
∑ s s (Ti −1 − Ti )
F C
cp C
k∈ΓiH s∈ΓiC
q
δn
TRANSSHIPMENT MODEL
Assume now that we do the same cascade for
si,k,1 each hot stream, while we do not cascade the
ri,1 pk,1 cold streams at all. In addition we consider
δi,1 si,k,2
heat transfer from hot to cold streams in each
ri,2 pk,2
interval.
The material balances for hot streams are:

si,k,j
ri,j pk,j δ i ,0 = 0
δi,j si,k,,j+1 δ i , j =δ i , j −1 +ri , j − ∑ si ,k , j ∀j = 1,...mI
k
ri,j+1 pk,j+1
The material balances for cold streams are:
δi,j+1
pk , j = ∑ si ,k , j ∀j = 1,...mI
i
si,k,n
ri,n pk,n
Where ri,j and pk,j are the heat content of hot
δi,n stream I and cold stream k in interval j.
TRANSSHIPMENT MODEL
si,k,1
Although we have a simpler model to solve it,
ri,1 pk,1 in this new framework, the minimum utility
problem becomes:
δi,1 si,k,2
ri,2 pk,2
Min δ U , 0
s.t
si,k,j δ i ,0 = 0 ∀i, ∀j = 1,...mI
ri,j pk,j
δ i , j =δ i , j −1 +ri , j − ∑ si ,k , j ∀i, ∀j = 1,...mI
δi,j si,k,,j+1 k
pk , j = ∑ si ,k , j ∀k , ∀j = 1,...mI
ri,j+1 pk,j+1
i
δi,j+1
Note that the set of hot streams now includes

si,k,n
ri,n pk,n process streams and the utility U. Cold
streams include cooling water.
δi,n
TRANSSHIPMENT MODEL
si,k,1 We would like to have a model that would tell
ri,1 pk,1 us the si,k,j such that the number of units is
minimum. We now introduce a way of counting
δi,1 si,k,2
matches between streams. Let Yi,k be a binary
ri,2 pk,2
variable (can only take the value 0 or 1).
Then we can force Yi,k to be one using the

si,k,j
ri,j pk,j
following inequality
δi,j si,k,,j+1
∑ j
si ,k , j − ΓYi ,k ≤ 0
ri,j+1 pk,j+1
δi,j+1 indicating therefore that heat has been

transferred from stream i to stream k in at least
si,k,n one interval.
ri,n pk,n
δi,n
TRANSSHIPMENT MODEL
si,k,1 The complete model would be:
ri,1 pk,1
∑∑
δi,1 si,k,2
Min Yi ,k
ri,2 pk,2 i k
s.t
δ U ,0 = δ U* , 0
si,k,j
ri,j pk,j δ i ,0 = 0 ∀i, ∀j = 1,...mI
δi,j si,k,,j+1 δ i , j =δ i , j −1 +ri , j − ∑ si ,k , j ∀i, ∀j = 1,...mI
k
ri,j+1 pk,j+1
pk , j = ∑ si ,k , j ∀k , ∀j = 1,...mI
δi,j+1 i
∑ si ,k , j − ΓYi ,k ≤ 0 ∀i, ∀k
si,k,n j
ri,n pk,n
δi,n The model can only be solved above and

below the pinch separately. Why???
TRANSSHIPMENT MODEL
We are minimizing the number of matches.

Different answers can be obtained if separate
regions are not considered. These answers are
not guaranteed to be realistic.
GAMS MODEL
TABLE P(K,J) load of cold stream K1 in interval J
Min ∑∑i k
Yi ,k J0 J1 J2 J3
C1 0 8.6 6.45 2.15
s.t C2 0 0 0 1
δ U ,0 = δ U* , 0 W 0 0 0 6.4 ;
δ i ,0 = 0 ∀i, ∀j = 1,...mI VARIABLES

S(I,K,J) heat exchanged hot and cold streams
δ i , j =δ i , j −1 +ri , j − ∑ si ,k , j ∀i, ∀j = 1,...mI D(I,J) heat of hot streams flowing between intervals
k
Y(I,K) existence of match
pk , j = ∑ si ,k , j ∀k , ∀j = 1,...mI Z total number of matches ;
i
∑ j
si ,k , j − ΓYi ,k ≤ 0 ∀i, ∀k POSITIVE VARIABLE S
POSITIVE VARIABLE D
BINARY VARIABLE Y ;
EQUATIONS
GAMS MODEL MINMATCH objective function-number of matches
HSBAL(I,J) heat balances of hot stream I in INTERVAL J
SETS CSBAL(K,J) heat balances of cold stream J1 in K
I hot streams above pinch / S, H1 /
HTINEQ1(I,K) heat transferred inequalities;
K cold streams above pinch / C1,C2,W/
J temperature intervals / J0*J3 / ;
MINMATCH .. Z =E= SUM((I,K), Y(I,K));
HSBAL(I,J)$(ORD(J) NE 0) .. D(I,J)-D(I,J-1)+ SUM(K,S(I,K,J)) =E= R(I,J);
SCALAR GAMMA /10000/; CSBAL(K,J)$(ORD(J) NE 0) .. SUM(I, S(I,K,J)) =E= P(K,J) ;
HTINEQ1(I,K) .. SUM(J, S(I,K,J))-GAMMA*Y(I,K) =L= 0 ;
TABLE R(I,J) load of hot stream I1 in interval K
J0 J1 J2 J3 MODEL TSHIP /ALL/ ;
S 9.2 0 0 0 SOLVE TSHIP USING MIP MINIMIZING Z;
H1 0 0 6.75 2.25; DISPLAY S.L, D.L, Y.L;
GAMS MODEL
SOLUTION
---- VARIABLE S.L

J1 J2 J3 550 0C
S .C1 8.600 0.600
H1.C1 5.850 2.150 750 0C
H1.C2 1.000
FCp=0.045 H1
---- VARIABLE D.L H2

FCp=0.04
J0 J1 J2
S 9.200 0.600
H1 0.900 900 0C
FCp=0.043
9.2 8
---- VARIABLE Y.L FCp=0.02 550 0C
C1 C2 1 500 0C
S 1.000
H1 1.000 1.000
EXECUTION TIME = 0.090 SECONDS

PART 4
UTILITY PLACEMENT
HEAT AND POWER
INTEGRATION
UTILITY PLACEMENT
We now introduce the GRAND COMPOSITE CURVE, which will be useful to
analyze the placement of utilities.
1.5
T
Start at the pinch
250
- 6.0
240
1.0 200
190
-4.0
150
Pinch
14
80
-2.0 40
30
-2.0
∆H
GRAND COMPOSITE CURVE
T Total heating utility
250
240
200
190 These are called
150 “pockets”
Process-to Process
80
integration takes
40
30 Total cooling utility place here
∆H
UTILITY PLACEMENT
We now resort to a generic grand composite curve to show how
utilities are placed.
T HP Steam
MP Steam
LP Steam
Cooling Water
∆H
Transshipment model for multiple utilities
Min ∑c
m∈S
m QmS + ∑ cnQnW
n∈W
s.t
δ i ,0 = 0 ∀i, ∀j = 1,...mI
δ i , j =δ i , j −1 +ri , j − ∑ si ,k , j − ∑ Qi ,n , j ∀i, ∀j = 1,...mI
k m∈S
δ m , j =δ m , j −1 − ∑ Qi ,n , j + QmS ∀i, ∀j = 1,...mI

m∈S
pk , j = ∑ si ,k , j + ∑ Qm ,k , j ∀k , ∀j = 1,...mI
i m∈S
∑Q
m∈S
m ,k , j = QnW ∀k , ∀j = 1,...mI
Results HOT UTILITY 1 = HP STEAM HOT UTILITY 2 = MP STEAM
Interval T( i ) Q( i ) δ( i ) HU1( i ) HU2( i ) CU1( i )
I0 250
I1 200 -4.5 0 4.5 HOT UTILITY 1
I2 150 -3 0 3 HOT UTILITY 2
I3 30 10 0 10 COLD UTILITY
UTILITY PLACEMENT
Hot Oil placement and extreme return temperatures
Hot oil from

T furnace
Oil minimum return

temperature
Water maximum Process

return temperature
Cooling water
∆H
UTILITY PLACEMENT
Furnace Theoretical Flame
Temperature
Tstack
Air
Tstack
Fuel Process
Stream
Ambient Temperature
Stack Loss
Fuel heat value ∆H
COMBINED HEAT AND POWER
Integration of a Heat Engine below the Pinch.
QH,min
T
T
QHE
Pinch
W Note that in this case
there is no gain. The heat
engine can be arranged
QHE -W separately and the utility
usage will not change.
QC,min +( QHE –W)

Integration of a Heat Engine Across the Pinch.
QH,min -( QHE –W) TOTAL HEN
QHE
ENERGY INTAKE
QH,min -( QHE –W)
T
W (Smaller)
QHE -W SYSTEM TOTAL

Pinch ENERGY INTAKE
QH,min+W
vs. QH,min+QHE
(separate)
QC,min
INTEGRATION WITH
DISTILLATION
Placement below the pinch.
QH,min
In this case there is a

Pinch
gain of (Qcond - Qreb)
in the cooling utility.
Qreb
Qcond
QC,min + (Qcond - Qreb)

HEAT PUMPS/REFRIGERATION
CYCLES
QH,min-(QHP + W) TOTAL HEN INTAKE
QH,,min – (QHP+W)
QHP +W (smaller)
TOTAL SYSTEM
Pinch
INTAKE
W QH,,min -QHP
vs. QH,,min +W (separate)
QHP
Savings in cooling utility

QHP
QC,min - QHP
Utility Placement
QHP
QH,min –( QHP+QLP)
T QHE T
QHP W QLP
Pinch QLP
TOTAL HEN INTAKE
QH,min –( QHP+QLP)
(smaller)
QC,min
TOTAL SYSTEM INTAKE
QC,min QH,min +W
vs. QH,min+W+(QHP + QLP) ∆H
(separate)
Gas Turbine Placement
TOTAL ENERGY INTAKE
TEX
QH,min + W+ QLOSS
vs. QH,min + W+ QLOSS+QS

T
QH,min -QS QF
T
W
Air
QS
Pinch TEX
T0
QLOSS QS
QLOSS
QF -W ∆H
QC,min T0
PART 5
DISTILLATION PLACEMENT
PLACEMENT OF DISTILLATION
Placement across the pinch.
QH,min + Qreb
T Qreb
Pinch
Note that in this case there is

no gain. The distillation
Qcond
column can be arranged
separately and the utility
usage will not change.
QC,min + Qcond
PLACEMENT OF DISTILLATION
Placement above the pinch.
QH,min + Qreb - Qcond
T Qreb
Qcond
Pinch Note that in this case
there is a possible gain
in the heating utility.
QC,min
ADJUSTING PRESSURE FOR
PROPER PLACEMENT
Heating Utility
T T
Pinch Pinch
As pressure increases utility usage decreases.

CRUDE FRACTIONATION
EXAMPLE
• We now show a complete analysis of an atmospheric crude
fractionation unit. We start with the supply demand diagram
CO NDENSER
w a te r
n a p h th a
PA1
ste a m
PA2 k erosen e
ste a m
PA3 d ie s e l
w a te r
cru d e DESALTER ste a m
HEN
g a s o il
so u r w a te r ste a m
HEN FURNACE r e s id u e
CRUDE FRACTIONATION
EXAMPLE
• We now show how to determine the heat load of pumparounds.
• We start with a column with no pumparound (results from are
from a rigorous simulation)
1.2 PINCH
1.0
M*Cp, MMW/C
0.8 COND
CRUDE
0.6
0.4
1
0.2 RES
0
0 200 300 400
PRODUCTS TEMPERATURE, °C
CRUDE FRACTIONATION
EXAMPLE
• We move as much heat from the condenser to the first
pumparound as possible. The limit to this will be when a
plate dries up. If the gap worsens to much, steam is added.
1.4
1.2 PINCH
M*Cp, MMW/C
1.0
0.8
PA1
0.6 CRUDE
COND
0.4
0.2
RES
0
0 100 200 300 400
PRODUCTS TEMPERATURE °, C
CRUDE FRACTIONATION
EXAMPLE
• We continue in this fashion until the total utility reaches a minimum. .
1.4
PINCH
1.2
PA1
1.0
M*Cp, MMW/°C
0.8 CRUDE
0.6
COND
0.4
PA2
0.2
RES
0
0 100 200 300 400
TEMPERATURE, °C
CRUDE FRACTIONATION
EXAMPLE
• Especially when moving heat from PA2 to PA3, steam usage increases
so that the flash point of products is correct and the gap is within
limits.
1.2
1.0
M*Cp, MMW/C
0.8
PA1
CRUDE
0.6
COND
0.4
1 PA2
PA1
PA3
0.2
RES
0
0 100 200 300 400
TEMPERATURE, °C
CRUDE FRACTIONATION
EXAMPLE
• Situation for a heavy crude
0.8
0.6 CRUDE PA3

M*Cp, MMW/C
0.4
RES
COND
0.2
SW PA1
PA2
0
0 100 200 300 400
TEMPERATURE, °C
PART 6
ENERGY RELAXED
NETWORKS
ENERGY RELAXATION
Energy relaxation is a name coined for the
procedure of allowing the energy usage to
increase in exchange for at least one of the
following effects :
a) a reduction in area
b) a reduction in the number of heat exchangers
c) a reduction in complexity (typically less
splitting)
ENERGY RELAXATION IN THE
PINCH DESIGN METHOD
• LOOP: A loop is a circuit through the

network that starts at one exchanger and
ends in the same exchanger
• PATH: A path is a circuit through the

network that starts at a heater and ends at
a cooler
ENERGY RELAXATION
• Illustration of a Loop
175 0C 125 0C
FCp=0.010 H1 C
280
FCp=0.040 H2
65 0C
155 0C 20 0C
FCp=0.020 C1
H
360 500 1840

40 0C C2
FCp=0.015
112 0C
560 520
(*) Heat exchanger loads are in kW ∆Tmin=13 oC

ENERGY RELAXATION
• Illustration of a Path
175 0C 125 0C
FCp=0.010 H1 C
280
FCp=0.040 H2
65 0C
FCp=0.020 155 0C 20 0C
H C1
360 500 1840

FCp=0.015
40 0C C2
112 0C
560 520
(*) Heat exchanger loads are in kW
ENERGY RELAXATION
• Procedure : Find a loop and move around heat from
exchanger to exchanger until one exchanger is eliminated.
175 0C 125 0C
H1 C
280
H2
65 0C
155 0C
20 0C
H C1
360 500 -X 1840 +X
40 0C C2
112 0C
560 -X 520 +X
• If one wants to eliminate one exchanger: X=500. Note that X

could have been negative, but we chose the smallest possible in
absolute value.
ENERGY RELAXATION
• Result: Notice that the result is infeasible!!!
73 0C
H1 175 0C 45 0C
C
65 0C 280
H2 65 0C
65 0C
155 0C 20 0C C1
H
360 2340
112 0C
40 0C C2
60 1020
108 0C
This exchanger is in
violation of the
minimum approach
ENERGY RELAXATION
• We use a path to move heat around to restore feasibility
73 0C
H1 175 0C 45 0C
C
65 0C 280 +X
H2 65 0C
65 0C
155 0C 20 0C C1
H
360 +X 2340 -X
112 0C 40 0C C2
60 +X 1020 -X
108 0C
The value of X needed to restore feasibility is X=795

ENERGY RELAXATION
• Final Network
H1 175 0C 45 0C
C
65 0C 1075
H2 65 0C
65 0C
155 0C 20 0C C1
H
1155 1545
112 0C
40 0C C2
855 225
55 0C
TEMPERATURE APPROACH (∆T)
RELAXATION
• We recall stating “NO HEAT ACROSS THE PINCH”.
• Being more specific, we should say, “NO NET HEAT ACROSS
THE PINCH”. Thus we allow the following situations.
Pinch
as long as the NET heat transferred across the pinch is zero.

However this implies allowing the temperature difference to be lower than
∆Tmin
∆T RELAXATION
• We thus define two types of Minimum Temperature Approach.
• HRAT: (Heat Recovery Approach Temperature): This is the

∆Tmin we use to calculate minimum utility.
• EMAT: (Exchanger Minimum Approach Temperature): This is

the minimum approach we will allow in heat exchangers.
When EMAT< HRAT networks can have

1. less splitting
2. less number of units
3. No significant increase in the total area.
∆T RELAXATION
Consider the following problem
Stream Type Supply T Target T F*Cp
(oC) (oC) (MW oC-1)
H1 Hot 175 45 0.010
H2 Hot 125 65 0.040
C1 Cold 20 155 0.020
C2 Cold 40 112 0.015
We now consider HRAT=20 oC and EMAT= 13 oC. The

corresponding minimum utility are:
∆Tmin Hot Utility Cold Utility Pinch

20 (HRAT) 0.605 MW 0.525 MW 132 OC
13 (EMAT) 0.360 MW 0.280 MW 112 OC
PSEUDO-PINCH METHOD
We now consider that the difference (245 kW= 605
kW-360 kW) needs to go across the pinch of a design
made using EMAT. Thus we first look at the solution
of the pinch design method (PDM) for ∆Tmin =13 oC
175 0C 125 0C 45 0C
FCp=0.010 H1
280
FCp=0.040 H2
65 0C
155 0C 20 0C C
FCp=0.020 H 1
360 500 1840

FCp=0.015 40 0C
C2
112 0C
560 520

PSEUDO-PINCH METHOD
To relax this network by 245 kW we extend the only
heat exchanger above the pinch by this amount. We
then proceed below the pinch as usual.
175 0C 149.5 0C 45 0C
FCp=0.010 H1 C
310
125 0C 65 0C
FCp=0.040 H2 C
215
155 0C 112 0C 20 0C
FCp=0.020 H C1
605 255 1840
FCp=0.015 40 0C
C2
112 0C
735 345
Note that the matching rules (FCp inequalities) can be somewhat relaxed.
PSEUDO-PINCH METHOD
We know the solution of the pinch design
method (PDM) for ∆Tmin =20 oC
175 0C 125 0C 45 0C
FCp=0.010 H1
525
125 0C 65 0C
FCp=0.040 H2
155 0C 20 0C C
FCp=0.020 H 1
105 0C
500 500 1700
FCp=0.015 112 0C 40 0C C
H 2
105 700 275
The PDM produces one additional split, and two

heaters, while the PPDM features two coolers
DIFFICULTIES IN THE PPDM
METHOD
• No clear indication what to do when there
is many hot streams above the pinch. How
to distribute the difference in heat?
• Even if the above is clarified it is not
practical for more than a few streams.
∆T RELAXATION USING
AUTOMATIC METHODS
• One can use the Transshipment model fixing the
level of hot utility and creating the intervals using
EMAT instead of HRAT.
• More sophisticated methods add area estimation to
the model (This has been called the Vertical model)
• The number of units can also be controlled and
sophisticated techniques have been used to explore
all the flowsheets with the same number of matches.
• Finally, once a flowsheet is obtained a regular
optimization can be conducted.
• This will be explored in the next Part.
PART 7
MATHEMATICAL
PROGRAMMING
APPROACHES
RECENT REVIEW PAPER
(until 2000)
• Furman and Sahinidis. A critical review and

annotated bibliography for Heat Exchanger
Network Synthesis in the 20th Century. Ind.
Eng. Chem. Res., 41 pp. 2335-2370, (2002).
PINCH DESIGN METHOD
• It is a DECOMPOSITION approach (3 steps)

– Perform Supertargeting and obtain the right HRAT,
the pinch (or pinches) and the minimum utility usage
– Pick matches away from the pinch using the tick-off
rule
– Evolve into higher energy consumption solutions by
loop breaking and adjusting loads on paths.
SUPERSTRUCTURE APPROACH
• CONCEPT: A single optimization model, if solved
globally, provides all the answers simultaneously.
Superstructure of matches.
Possible flow sheets embedded (recycles/by-passes excluded)
Possible flow sheets embedded (continued)
Model Constraints
Objective (no fixed costs)

– This is an MINLP formulation with which several
researches have struggled. (MINLP methods could
not be easily solved globally until recently (?).
– Therefore it needs some initial points.
ALTOUGH A ONE-STEP CONCEPT IT BECAME

IN REALITY AN ITERATIVE PROCESS
• Some Strategies to overcome the “curse of
non-convexity”
Hasemy-Ahmady et al.,
1999.
Many other methodologies attempted this goal (provide good initial

points) like evolutionary algorithms, simulated annealing, etc.
STAGE SUPERSTRUCTURE
APPROACH
• With isothermal mixing. Yee and Grossmann
(1990, 1998) Note on the side: This
is a remarkable coming
back to origins.
From Grossmann and Sargent (1978)
In the 90’s the mathematical programming/ superstructure

approach emerged as the dominant methodology
From Grosmann and Sargent (1978)

LATEST MILP APPROACH
(Barbaro and Bagajewicz, 2005)
Counts heat exchangers units and shells

Approximates the area required for each exchanger unit or shell
Controls the total number of units
Implicitly determines flow rates in splits
Handles non-isothermal mixing
Identifies bypasses in split situations when convenient
Controls the temperature approximation (HRAT/EMAT or ∆Tmin)
when desired
Allows multiple matches between two streams
m 1 2 3 4
5 6 Hot stream i
7 8
z,H
qîjm
qimz , jn
z ,C
qîjn
Cold stream j
n 1 2 3 4 5
6 7 8 9
Transportation Model Approach

Heat Exchanger counting m
z, H
Yijm z,H
K ijm
z,H
K̂ ijm
Kˆ ijm
z ,H
≥ Yijmz , H First interval
1 0 0 0
2 0 0 0
Kˆ ijm
z ,H
≤ 2 −Yijm
z ,H
−Yijm
z ,H
+1
3 1 1 0
4 1 0 0
Kˆ ijm
z ,H
≤ Yijm
z ,H
Rest of intervals 5 1 0 0
Kˆ ijm
z ,H
≥ Yijm
z ,H
−Yijm
z ,H
+1
6 1 0 0
7 1 0 0
Kˆ ijm
z ,H
≥0 8 1 0 1
9 0 0 0
10 0 0 0
m 1 2 3 4 5 6 7 8 9 10
Flowrate Consistency
z ,H z ,H z,H z,H
q̂ijm q̂ijm −1 qîjm qîjm−1
≥ ≤
Cpim( TmU − TmL ) Cpim−1( TmU−1 − TmL−1 ) Cpim (TmU − TmL ) Cpim−1 (TmU−1 − TmL−1 )
Heat exchanger spanning
m 1 2 3 4
6 7
8 9
z,H z,H
qîjm qîjm−1
=
Cpim (T − T )
U
m m
L
Cpim−1 (TmU−1 − TmL−1 )
PART 8
INDUSTRIAL IMPORTANCE
OF USING THE RIGHT MODEL
Crude fractionation case study

CRUDE FRACTIONATION
EXAMPLE
CO NDENSER
w a te r
n a p h th a
PA1
stea m
PA2 k erosen e
ste a m
PA3 d ie s e l
w a te r
cru d e DESALTER stea m
HEN
g a s o il
so u r w a te r ste a m
HEN FURNACE r e s id u e
CRUDE FRACTIONATION EXAMPLE
Light Crude Heavy Crude
400
400
350 350
300 300
Temperature ( C)
T emp eratu re (C)

o
250 250
200 200
150 150
100 100
50
50
0
0
0 50 100 150 200
0 50 100 150 200 250
Enthalpy (MW) Enthalpy (MW)
Cooling water is already included in the graphs.
•The light crude exhibits what is called a continuous pinch.

•The heavy crude is unpinched.
CRUDE FRACTIONATION
C2
Cooling water
FURNACE
H8
H2 H3 H7
H6
H5
H1
DESALTER
H9 H10
H4
MER network for the light crude.

CRUDE FRACTIONATION
C2
Cooling water
FURNACE
H8
H7
H6
H2
DESALTER
H5
H4
H3 H1
H10
C1
MER network for the heavy crude.

CRUDE FRACTIONATION
NOT USED FOR
C2
HEAVY CRUDE
FURNACE
NOT USED FOR
LIGHT CRUDE
Cooling water
H8
H2 H3 H7
H6
H5
H1
DESALTER
H4
H10
H9
C1
MER network efficient for both crudes.

CRUDE FRACTIONATION
It is clear from the previous results that efficient MER
networks addressing multiple crudes can be rather complex
and impractical.
Alternatives to the Pinch Design Method

(PDM) are clearly needed.
This does not mean that the PDM fails all the time. It is still
capable of producing good results in many other cases.
CRUDE FRACTIONATION
EXAMPLE
We now illustrate the use of HRAT/EMAT procedures for the
case of crude fractionation units.
We return to our example of two crudes. The problem was

solved using mathematical programming. The networks have
maximum efficiency for both crudes.
Only the vertical model and the control on the number of units
was used. No variation in the matches was done (not really
necessary in this case) and no further optimization was
performed.
CRUDE FRACTIONATION
C2
NOT USED FOR
HEAVY CRUDE FURNACE
Cooling water
H8
H7
H3 H6
H2
18 units H5
DESALTER
H1
NOT USED
FOR HEAVY H10
CRUDE
H4
H9
C1
Solution for HRAT/EMAT = 20/10 oF

CRUDE FRACTIONATION
NOT USED FOR
HEAVY CRUDE FURNACE
Cooling water
H8
H7
H3 H6
H2
H5
18 units
DESALTER
H1
NOT USED FOR
HEAVY CRUDE H10
H4
H9
C1

CRUDE FRACTIONATION
FURNACE
NOT USED FOR
HEAVY CRUDE
Cooling water
H8 H7
H3
H6
H2
DESALTER
17 units
H1 H5
H4
NOT USED FOR

HEAVY CRUDE
H10
H9
C1

CRUDE FRACTIONATION
Cost, MM$/yr
Combined Multiperiod Multiperiod+ Multip+Des.Temp
Network Model Desalt.Temp. +Higher HRAT
HRAT/EMAT 20/20 20/10 20/10 40/30
Operational 3.96 3.96 3.96 4.94
Fixed 3.14 3.63 3.32 1.37
Total 7.10 7.59 7.28 6.32
CRUDE FRACTIONATION
EXAMPLE
We now illustrate the use of more sophisticated models
that allow the control of splitting. These are essentially
transshipment models that are able to control the level of
splitting (something that regular transshipment models
cannot do).
CRUDE FRACTIONATION
Two branches unrestricted
HRAT = 40 oF FURNACE
EMAT = 30 oF H8
H7
H3
H6
23 units H2
H5
H4
DESALTER
5% More Energy H1
Consumption H10
H9
C1
CRUDE FRACTIONATION
restricted
HRAT = 40 oF FURNACE
EMAT = 30 oF
H7
H8
H3
H6
H2
H5
H1
21 units
H4
DESALTER
5% More Energy
H10
Consumption
H9
C1
CRUDE FRACTIONATION
Cost, MM$/yr
Multiperiod Two-branch Two-branch

Model Unrestricted Restricted
Operational 4.32 4.52 4.53
Fixed 2.07 1.90 2.01
Total 6.39 6.42 6.54
CRUDE FRACTIONATION
• The two-branch design…
– Is efficient for all feedstocks proposed.
– Consumes only a few more millions Btu/hr
– Has many solutions of similar energy consumption.
• Complexity can be reduced at a relatively small
energy increase and with some reduction of
capital.
• All these models suggest that there is a lot of
flexibility to perform an effective retrofit because
there are these many options of similar cost to
explore.
PART 9
RETROFIT
AND
TOTAL SITE INTEGRATION
RETROFIT
The big question in trying to do a retrofit of a HEN is
whether one really wants to achieve maximum efficiency.
Original System
Operating
Costs ($/yr)
HORIZON
Capital Investment ($)
Usually retrofits are too expensive and have a long payout.

RETROFIT
It is desired to produce the largest reduction of cost with
the smallest capital investment
Present Condition
Retrofits
HORIZON
Capital Investment ($)
The question is how to identify the most profitable.

EXAMPLE
Retrofits involve a) relocation, and b) addition of new units.
NEW-1
DESALTER NEW-2 9T-2

9T-1
NEW-3
REB-T7
9T-7
RELOC
NEW-4
9T-9 9T-1
RELOC NEW-5
9T-9
DESALTER
D101 9T-7
This particular study produced a) 700,000 annual savings with 1.2 years
payoff, b) Additional 12% capacity (not counted in the $700,000 savings)
TYPES OF RETROFIT
• By inspection. Perform pinch design or pseudo pinch
design and determine heat exchangers to add
• Systematic methods using tables and graphs exist
They are outside the scope of this course.
• Mathematical programming approaches also exist but
they have not passed the test of usability and
friendliness
RECENT WORK ON RETROFIT
• Asante, N. D. K.; Zhu, X. X. An Automated and Interactive Approach for Heat
Exchanger Network Retrofit. Chem.Eng. Res. Des. 75 (A), 349-360 (1997).
• Briones, V.; Kokossis, A. C. Hypertargets: A Conceptual Programming Approach

for the Optimisation of Industrial Heat Exchanger Networks II. Retrofit Design.
Chem. Eng. Sci. 54, 541-561 (1999).
• Barbaro, Bagajewicz, Vipanurat, Siemanond. MILP formulation for the Retrofit of
Heat Exchanger Networks. Proceedings of Pres 05. (2005)

• USE OF GRAND COMPOSITE CURVES TO PLACE UTILITIES.
Plant 1 Plant 2
∆H ∆H
Pockets are eliminated and curves are shifted

• Site sink and source profiles are constructed.
LP Steam HP Steam
Sink Profile
Source Profile
Overlap region. These are savings.
∆H
Energy integration is performed by placing utilities. We will

see how this method can be wrong.
• Consider two plants. We would like to know under what
conditions one can send heat from one plant to the other.
C O M B IN E D
PLANT 1 PLANT 2 PLANT
A b o v e b o th
p in c h e s
P in c h P o in t
P la n t 2
P o s s ib le
B e tw e e n
L o c a t io n o f
p in c h e s
P in c h
P in c h P o in t
P la n t 1
B e lo w b o th
p in c h e s
T U1,min- QT U2,min
Effective integration
takes place between
pinches.
Pinch 2
QT
From where plant 2 is
Pinch 1 heat source to where
plant 1 is heat sink.
W1,min W2,min- QT
T U1,min- QT U2,min + QT
QT Integration outside
Pinch 2 the inter pinch region
leads to no effective
savings.
Pinch 1
W1,min W2,min
GRAND COMPOSITE
CURVES
PLANT 1 PLANT 2
18 = 30 - 12 20 = 20
-12 -19
6 = 18 - 12 1 = 20 - 19
-1 -1
5=6-1 Pinch 0=1-1
-15 10 10
0 = 5 - 15 + 10 0 = 10 - 10
-2 2 5
Pinch 0=0-2+2 3=0+5-2
2 1
2=2 4=3+1
GRAND COMPOSITE CURVES
140
120 Plant
1
Temperature
100 Plan
t2
80
60 Maximum
possible savings
40
-20 -15 -10 -5 0 5 10 15 20
Enthalpy
ASSISTED HEAT TRANSFER
PLANT 1 PLANT 2
7 = 20 + 4 - 17 16 = 20 - 4
-7 -10
0=7-7 6 = 16 - 10
If heat is not transferred
above the two pinches 5 4 -10
only 13 of the maximum 1 = 0 + 5 - 4 Pinch 0 = 6 - 10 + 4
17 can be saved. -15 14 14
0 = 1 - 15 + 14 0 = 14 - 14
-3 3 10
Pinch 0=0-3+3 7 = 0 + 10 - 3
3 5
3=3 12 = 7 + 5
140
Pla
nt 1
120
Assisted
Temperature
Transfer Pocket
100
Plan
t2
80
Maximum
60
possible savings
40
-10 -5 0 5 10 15 20 25 30
Enthalpy
140
Pla
120 nt 1
Temperature
Typical solutions call for 100
“sealing the pocket” Plan
t2
preventing thus the savings. 80
Maximum
60 possible savings
40
-10 -5 0 5 10 15 20 25 30
Enthalpy
140
Pl a
nt 1
120
Assisted
Temperature
Transfer Pocket
100
P lan
t2
80
Maximum
60
possible savings
40
-10 -5 0 5 10 15 20 25 30
Enthalpy
HEAT EXCHANGER NETWORKS
Heat Exchanger networks should be such that both plants can work at maximum
efficiency when integrated and when they stand alone.
We now show a result of a case study. An integration between a Crude unit (heat sink )
and an FCC unit (heat source).
H9 H4 H3 H7 H6 H2 H1
H5
H8
C1
CW
Above the Pinch
Plant 1: Crude Unit H10 H11 H15 H13 H14
Direct Integration
(two stream circuits)
C2
Pinch location H12
CW
Original HEN
Additional heat exchangers Below the Pinch
used during integration
Plant 2: FCC Unit
HEAT EXCHANGER NETWORKS
Heat Exchanger networks should be such that both plants can work at
maximum efficiency when integrated and when they stand alone.
We now show a result of a case study. An integration between a Crude
unit (heat sink ) and an FCC unit (heat source).
H9 H4 H3 H7 H6 H2 H1
H5
H8
C1 Utility (MMBtu/hr)
Heating Cooling
CW No Integration 252.9 147.7
Above the Pinch Direct Integration 201.4 96.2
Plant 1: Crude Unit H10 H11 H15 H13 H14
Savings 1,000,000 $/year

Direct Integration
(two stream circuits)
C2
Pinch location H12
CW
Original HEN
Additional heat exchangers Below the Pinch
used during integration
Plant 2: FCC Unit
Multiple plants can also be analyzed. We show below one example of
such studies.
•Alternative solutions exists.
•Grand composite curves cannot be used anymore.
PLANT 1 PLANT 2 PLA N T 3 PLA N T 4
( T e s t C a s e # 2 ) ( T r iv e d i) (C & F ) (4 sp 1 )
0 2 5 4 .9 4 2 6 .4 128
300 °C
249 °C
9 8 .3
1 2 0 .7
5 2 .9
200 °C
1 0 4 .5
160 °C
1 0 7 .5
90 °C
40 °C
40 5 8 1 .1 1 9 9 5 .5 3 1 .1
FUTURE TRENDS
• Mathematical programming will be the dominant
tool.
• Software companies are struggling to make the
proper choice of existing methods for each case
and most important numerically reliable.
• The best option for the time being is to
intelligently interact with experts while using
existing software.

BagajewiczHeatIntegration PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

BagajewiczHeatIntegration PDF

Uploaded by

Copyright:

Available Formats

PART 1

PINCH AND MINIMUM

TC,in TC,out T TH,in

TH1,in TH1,out TH1,out

T • In general, the smallest ∆T

Stream Type Supply T Target T ∆H F*Cp

Note: There is a particular overlap that requires only cooling utility

Stream Type Supply T Target T ∆H F*Cp

1.5 1.5 This is the

1.0 1.0 This is the

DO NOT TRANSFER 1.5 Heating utility is

TAKES PLACE -2.0

TOTAL AREA TARGETING

The above scheme of heat transfer is called VERTICAL HEAT TRANSFER

Cooling 50 ∆T= ∆Tmin

Interval Q TH1 TH2 TC1 TC2 ∆Tml A

Total Area 2716.3

We will see later how the number of units can be

• Is the total area predicted this way, realistic?

YES, Within 10-15%

The area obtained is actually the minimum

Reactor 1 feed Cold 140 180 8.0 0.2

This implies at least three heat exchangers,

40 80 90 150 200 250 300

Consider the following warehouses, each containing some merchandise

The general answer is N=S-1 . When does one need less?

The general answer is N=S-P . P is the number of

Nmin= (S-P)above pinch+ (S-P)below pinch

If we do not consider two separate problems, above

• Heat exchangers will not contain heat

∆Tmin ∆Tmin Violation when FCpH > FCpC

Thus replacing one obtains

Q TC,in Golden rule for pinch

Violation when FCpC > FCpH

START BY MAKING PINCH

• Once a match has been selected how

THIS IS CALLLED THE

Stream Type Supply T Target T ∆H F*Cp

Stream Type Supply T Target T ∆H F*Cp

• Which matches are possible?

• The rule is that FCpH < FCpC . We therefore can only

• The tick-off rule says that a maximum of 8 MW is exchanged

• Which matches are possible?

FCp=0.2 140 0C 52.5 0C 20 0C

• The tick-off rule says that a maximum of 17.5 MW is

• Streams with unfulfilled targets are colored.

• Away from the pinch, there is more flexibility to make

230 0C 8 17.5 6.5

230 0C 8 17.5 6.5

Nmin= (S-P)above pinch+ (S-P)below pinch =

What is then, the solution?

Notice that different combinations of flowrates in the split

FCp=0.2 H1 150 OC 100 OC

Yes FCpH≤FCpC Yes

Place Split Hot

Yes FCpH ≥ FCpC Yes

The material balances for hot streams are:

Note that the set of hot streams now includes

Then we can force Yi,k to be one using the

δi,j+1 indicating therefore that heat has been

δi,n The model can only be solved above and

We are minimizing the number of matches.

δ i ,0 = 0 ∀i, ∀j = 1,...mI VARIABLES

Counts heat exchangers units and shells