1. Heat Exchanger Network Retrofit
Trevor Hallberg
Sarah Scribner

2. Heat Exchanger Network (HEN) Retrofit
Outline
Mixed Integer Linear Programming (MILP)
Pinch Technology Theory for Retrofit
Improvements on Pinch Technology
Crude Distillation Unit Example
Discussion

3. Heat Exchanger Network (HEN) Retrofit
Small Conceptual Example H1 C1 C2 1 5 4 HU CU
170°C
200°C
10 kW/°C
9 kW/°C
11 kW/°C
200˚C
30˚C
25˚C 2 3
55 kW
1000 kW
925 kW
420 kW
1110 kW
109.1˚C
76.7˚C
128˚C
35.5˚C
Retrofit Options :
Adjust existing area
Relocate existing exchangers
Add new exchangers
Introduce stream splits
Optimal Results :
Reduce utility usage
Increase process-process exchange
Maintain network integrity

4. Mixed Integer Linear Programming (MILP)
Based on transportation-transshipment model
Hot stream, i
Cold stream, j 3 4 5 6 7 8
Heat transfer Zones
Temperature Intervals
Energy and flow balances

5. Mixed Integer Linear Programming (MILP)
Heat Transfer Zones
Two Zones
Pinch Technology (Above and Below the Pinch)
One Zone
Find minimum total cost, even if cross-pinch transfer occurs

6. MILP Sophisticated cost analysis Model parameters
Area adjustment
Existing heat exchangers
New heat exchangers
Re-piping
Ability to tailor-fit model for a variety of HEN scenarios

7. Area Adjustments Area Reduction Area Addition Plug Tubes By-pass fluid
Enforce realistic area adjustments
Area reduction ≤ 50%
Area Addition
Existing Exchangers
Increase area of existing shell
Install a new, larger shell
New Exchangers
Limit number of new exchangers
Limit area (size) of new exchangers
Enforce realistic area adjustments
Area addition ≤ 20%

8. Re-piping Re-piping scenarios Re-piping cost assignments
Exchanger relocation
Stream splitting
Exchanger by-pass
Re-piping cost assignments
Model compares number of split streams in retrofitted network to original network
Assigns user-defined fixed cost to number of new splits

9. MILP Objective Functions Maximize value of savings
Value of Savings = Utility Cost Savings – Annual Capital Cost
Maximize net present value
NPV = ∑(DiscountFactori * Utility Cost Savingsi ) – Capital Cost
Maximize return on investment
ROI = Annual Utility Cost Savings / Capital Cost

10. MILP INPUT Parameters OUTPUT Data Stream data (F·Cp)
Stream temperatures (Inlet & Target)
Cost functions
OUTPUT Data
Optimized objective function
Exchanger locations
Cost requirements
Utility savings

11. Pinch Technology The root of heat integration technology
Systematic methodology for retrofitting
Identifies locations where process change will reduce the overall energy consumption
Allows us to set these energy targets
Allows us to set total cost targets

12. Pinch Technology Based on thermodynamic principles
Uses Temperature vs. Enthalpy diagrams
Called composite curves
HCC profiles the heat availability in the process
CCC profiles the heat demands in the process
combinations of straight line representations of the hot and cold streams based on the heat capacity flow rate (CP = FCp)

13. Pinch Technology Composite Curves - Combined
Displays the process minimum approach temperature (ΔTmin)
Displays maximum process-to-process heat recovery
Displays minimum utility requirements
Minimum hot utility = QH,MIN
Minimum cold utility = QC,MIN

14. Pinch Technology The pinch (ΔTmin)
Splits the process into 2 regions that are analyzed separately
“Heat Sink” region above the pinch
“Heat Source” region below the pinch
Quantifies how close the composite curves can be without violating the second law of thermodynamics
2nd Law: no temperature crossover within any of the enthalpy intervals

15. Pinch Technology 3 Pinch Rules
No heat transfer across the pinch
No external cooling above the pinch (only HU)
No external heating below the pinch (only CU)
Violating these results in cross-pinch heat transfer
Increases heat requirements

16. Pinch Technology Total Network Area - Aideal Q = U•Ainterval•ΔTLM
“Vertical Heat Transfer”- vertical enthalpy regions
Assumes equal area of each exchanger
Vary ΔTmin
Aideal = A1+A2+…+Ai
Q = heat transferred
U = overall heat transfer coefficient
ΔTLM = log mean temperature difference of the enthalpy intervals
A1 and A5 represent the areas of the utility exchangers
These will change as our process utility requirements change
Provides the minimum total

17. Pinch Technology Retrofitting Energy vs. Area diagram
Blue curve = Aideal for various ΔTmin values
Want to improve ineffective use of area
Want to decrease energy requirements
Optimum grassroots design curve assumes vertical heat transfer for the HEN
Total network area is calculated using the vertical heat transfer method described earlier
We have a few options for retrofitting from our existing design

18. Pinch Technology Retrofitting continued
Why would we increase area and energy?
Could theoretically work but the point is to use area better and decrease energy
Pinch recommends not decreasing area in which we have already invested - ???
For now we will assume this is our retrofit path
If we are going to increase area, why not try to decrease the energy at the same time? (blue)
Do not want to decrease area because we have already invested in it (pink)
Limitation but we will assume it is correct for now and will look into it more later (green)

19. Pinch Technology Need a way to compare energy and area
Want a curve similar to the optimum design curve
Need a way to determine the most economical solution on the new curve
We use something called area efficiency
Need a way to compare energy and area
Which path do we choose?

20. Pinch Technology Area Efficiency (α)
Based on utilities of current process
Agrassroots = optimal area for current process
Assume new design has αnew ≥ αcurrent
The best α can be is 1 (cannot be better than ideal)
Agrassroots is calculated using the vertical heat transfer method as described earlier for the current process
The higher αcurrent, the closer Aexisting is to Agrassroots

21. Pinch Technology Area Efficiency Assume α = β and
Aexisting
Aretrofit
Aideal
Agrassroots
Can now calculate Aretrofit based on constant curve
Process:
ΔTmin
Qu,min
Aideal
Aretrofit

22. Pinch Technology Which α do we use? Infinite amount of α values
At least want αcurrent
The best is α = 1
Larger α = smaller Aretrfit
Assume value of 1

23. Pinch Technology Now that we have Aretrofit?
We need the optimum ΔTmin value
Total Annualized Cost (TAC) vs. ΔTmin diagram for constant α
Optimum ΔTmin value corresponds to the minimum

24. Pinch Technology TAC vs. ΔTmin Nmin = [Nh+Nc+Nu-1]AP + [Nh+Nc+Nu-1]BP
Still assuming that the total area of the network is distributed evenly among the exchangers

25. Pinch Technology Now we need to design the HEN
Eliminate cross-pinch heat exchangers (E2)
Reuse the other exchangers (usually more economic)
Design sections above and below pinch separately H1 C1 C2 5 4 HU CU
170°C
200°C
10 kW/°C
11 kW/°C
9kW/°C
200˚C
30˚C
25˚C 1 2 3
55 kW
1000 kW
925 kW
420 kW
1110 kW
109.1˚C
76.7˚C
128˚C
35.5˚C
40˚C
Design is based on knowing the optimum ΔTmin value
maximizing the heat load for each exchanger ensure the minimum number of units
Designing above the pinch separately from below the pinch ensures no cross-pinch heat transfer
Match one hot and one cold stream per exchanger

26. Pinch Technology HEN Design Start design at pinch
Matching streams
AP: (F•Cp)hot ≤ (F•Cp)cold
BP: (F•Cp)cold ≤ (F•Cp)hot
Maximize exchanger loads
Q = F•Cp•ΔT (for each stream) H1 C1 C2 1 5 4 HU CU
170°C
200°C
10 kW/°C
11 kW/°C
9 kW/°C
200˚C
30˚C
25˚C 2 3
45 kW
1870 kW
55 kW
1300 kW
230 kW
109.1˚C
76.7˚C
128˚C
35.5˚C
40˚C

27. Pinch Technology HEN Design Add E6 to reduce utilities
Use loops and paths to make design more flexible
Give E6 a duty of X
A web of exchangers is affected H1 C1 C2 1 5 4 HU CU
170°C
200°C
10 kW/°C
11 kW/°C
9kW/°C
200˚C
30˚C
25˚C 2 3
45 kW
X kW
55 kW
X kW
230 + X kW
109.1˚C
76.7˚C
128˚C
35.5˚C
40˚C 6
Loops give the process a flexibility can be used to make old exchangers fit new duties
A path gives flexibility also and is a connection through streams and exchangers between two utilities
Loops and paths allow us to shift heat loads in a network
X kW

28. Pinch Technology Individual Heat Exchanger Area
Evaluate the specific exchanger areas by accounting for temperature cross within the exchangers
Need these areas to calculate the capital cost
Q = heat load on the exchanger
U = overall heat transfer coefficient (based on the two matched streams)
ΔTLM = log mean temperature difference of the exchanger

29. Pinch Technology Cost comparisons
No longer assume equal areas for each exchanger
CC (Capital Investment Cost)
Based on area change for each exchanger in the network
Fixed and variable costs
Includes area addition, reduction, and new exchanger cost
Operating Costs (OC)
∆TAC (Total Annualized Cost)
ROI (Return on Investment)

30. Pinch Technology How can we improve it?
Allowing relocation of all exchangers
May be able to cut down on area change expenses
May decrease the number of new exchangers needed
Incorporate Pro-II simulation

31. Pinch Technology Improvement
Incorporate Pro-II Optimization
Pro-II simulation based on Pinch Technology results
Exchanger location
Minimize total cost
Vary heat exchanger area
Vary stream split ratio
Fix stream target temperatures

32. Pro-II Optimization Controllers set stream target temps
Calculator assigns cost equations
Optimizer minimizes cost function

33. HEN Retrofit Results Crude Distillation Unit Retrofit Considerations
MILP
Process Pinch
Process Pinch Improvements
Pro-II Simulation
Retrofit Considerations
Stream splitting
Addition of new exchangers
Allow & disallow exchanger relocation

34. Crude Distillation Unit
Original Network
10 Hot streams
3 Cold streams
18 exchangers
2 hot utilities
3 cold utilities

35. Original MILP Process Pinch
No Relocation Allowed
Original MILP Process Pinch
COMPARISON
Original
18 exchangers
MILP
8 new exchangers
Process Pinch
9 new exchangers
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1

36. Retrofit Results MILP has 1 less exchanger
MILP has more area but requires less energy
Pinch has less area but requires more energy
Pro-II simulation further optimizes the pinch technology 36

37. Original MILP Process Pinch
Allow Relocation
Original MILP Process Pinch
COMPARISON
Original
18 exchangers
MILP
5 new exchangers
5 relocated
Process Pinch
9 new exchangers
7 relocated
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1 37

38. Retrofit Results
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1 38

39. Discussion Computation Time Comparison MILP is most time-efficient
MILP only requires input of data
Pro-II requires large majority of manual labor
Pinch requires manual labor only

40. Where Does Pinch Go Bad? We believe the problem is with ΔTmin
The optimum value is determined prior to design
Assumes equal area of every exchanger
Optimization occurs after the value is chosen

41. MILP = The Best Why? Considers the greatest number of variables
Considers all solutions
No limiting assumptions or methodology
Optimization based on several cost parameters
Computer does everything a person can do
Only requires input of data
Do not need experience with the methodology

42. The End

43. Process Pinch Limitations
Why does pinch overlook MILP solutions?
Optimum solution based on min number of units
It then optimizes the area distribution and utilities
For large networks, there are many (if not infinite) combinations of exchangers and heat loads to satisfy a process
Cannot efficiently consider all heat loops and paths

44. Example 1 Original Network 7 exchangers 1 heater (E7)
2 coolers (E5,E6)
Stream F
kg/s Cp
kJ/kg.C
Tin oC
Tout H
kW/m2.oC H1
228.5 1
159 77
0.4 H2
20.4
267 88
0.3 H3
53.8
343 90
0.25
HU (hot utility)
500
499
0.53 C1
93.3 26
127
0.15 C2
196.1
118
265
0.5
CU (cold utility) 20 40 H1 H2 H3 C1 C2 3 5 1 4 2 6 7 HU CU
159 0
127 0
228.5
20.4
53.8
93.3
196.1
267˚
343˚
265˚
77˚
88˚
90˚
26˚
118˚

45. Example 1- Process Pinch
Retrofitted Network (No relocation Allowed)
8 exchangers (E6 becomes non-operational)
1 heater (E7), 1 cooler (E5)
E8 and E9 added to increase process exchange H1 H2 H3 C1 C2
159 0
127 0
267˚
343˚
265˚
77˚
88˚
90˚
26˚
118˚
New -A +A
NEW SPL 1 4 2 3 8 9 5 HU 7 CU 6
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1

46. Example 1 – Process Pinch
Retrofitted Network (Relocation Allowed)
8 exchangers (E6 is relocated a process exchanger)
1 heater (E7), 1 cooler (E5)
E8 added to meet increase process exchange H1 H2 H3 C1 C2
159 0
127 0
267˚
343˚
265˚
77˚
88˚
90˚
26˚
118˚
New -A +A
NEW SPL 1 2 3 8 6 4 5 HU 7 CU +A
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1 46

47. Example 1- MILP Retrofitted Network (No relocation Allowed)
9 exchangers
1 heater (E7), 2 coolers (E5, E6)
E8,E9 added to increase process exchange
77˚ H1 H2 H3 C1 C2
88˚
90˚
26˚
118˚
267˚
343˚
265˚
159˚
127˚
New
+A , NS
NEW SPL -A 3 4 9 1 2
+A, NS 5 6 HU CU 7 8
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1

48. Example 1- MILP Retrofitted Network (Relocation Allowed) 9 exchangers
1 heater (E7), 2 coolers (E5, E6)
E8,E9 added to increase process exchange
77˚ I1 I2 I3 J1 J2
88˚
90˚
26˚
118˚
267˚
343˚
265˚
159˚
127˚ 5
New +A
+A, NS
NEW SPL 3 1 2 10 HU CU -A 7 6 9 8 4
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1

49. EXAMPLE 1 RESULTS BEYOND THIS POINT

50. Example 1 Results (No relocation) MILP has 1 more exchanger than pinch
MILP has more area but requires less energy
Pinch has less area but requires more energy
Pro-II simulation further optimizes the pinch technology
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1 50

51. Example 1 Results (Relocation allowed)
MILP has 2 more exchangers than pinch
MILP has more area but requires less energy
Pinch has less area but requires more energy
Pro-II simulation further optimizes the pinch technology
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1 51

52. Example 1 Results (No Relocation)
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1 52

53. Example 1 Results (Relocation Allowed)
E9 satisfies the heat requirement for hot stream 3 below the pinch temperature and so E6 is not needed as a utility exchanger
a split is needed on cold stream 1 53