Sabp A 009

Best Practice
SABP-A-009 Pinch Technology for Energy Efficiency Optimization Document Responsibility: CSD/ESD/Energy Systems Unit 9 May 2006
Pinch Technology for Energy Efficiency Optimization
Developed by: Energy Systems Unit Consulting Services Department Issue Date: 22 April 2006
Previous Issue: None Next Planned Update: 9 May 2009 Page 1 of 208 Primary Contact: nourelmm@aramco.com, phone +966 (3) 873-6045
Document Responsibility: Energy Systems Unit, CSD Issue Date: 9 May 2006 Next Update: 9 May 2009
SABP-A-009 Pinch Technology for Energy Efficiency Optimization
Table of Contents
1.0
Introduction 1.1 1.2 1.3 Definition Purpose and Scope Intended Users
3 3 3 3
2.0
Pinch Technology 2.1 2.2 2.3 2.4 2.5 History and Background Data Extraction Targeting Heat Exchangers Network (HEN) Design Heat Exchangers Network (HEN) Improvement
4 4 22 31 114 169
3.0
Pinch Technology and Mathematical Programming Hybrid Methods 3.1 3.2 Mathematical Programming for HEN Synthesis Simultaneous Process Optimization and Heat Integration
173 178 192
4.0
Appendices 4.1 4.2 Computer Tools References
207 207 207
Page 2 of 208
1.0
Introduction
Energy conservation in Saudi Aramco became everyones business. It is mandatory for each process facility to find cost effective solutions to save energy and achieve more with less in their facilities. Saudi Aramco has constituted a committee called EMSC Energy Management Steering Committee to direct and manage a sustainable process for energy conservation. A vital contribution towards the success of the company wide energy conservation policy comes through documenting the company best practices in methodology, tools and applications in the field of energy conservation and distributing such knowledge among our facilities. Hence, a consistent effort has been exerted in Saudi Aramco to produce Best Practices to help Saudi Aramco plants achieve their energy conservation targets and assimilate energy conservation knowledge. This particular Best Practice document for pinch technology is a contribution towards this goal. 1.1 Definition
The term pinch technology or sometimes called pinch analysis refers to the methodology of collecting and analyzing available process data to conduct what was originally called heat integration. Pinch technology is the technology that provides a systematic methodology for energy saving in processes and total sites. The methodology is based upon thermodynamic principles. 1.2 Purpose and Scope
The purpose of this best practice document is to introduce the pinch technology short cut tools by which energy integration and heat exchanger networks synthesis and retrofit can be conducted faster, cheaper and better in our facilities. Its scope include the pinch technology tools for energy integration and heat exchangers network design/synthesis and the pinch-mathematical programming hybrid methods for simultaneous optimization of process conditions and heat integration. 1.3 Intended Users
This Best Practice manual is intended for use by the energy engineers working in Saudi Aramco plants, who are responsible for energy integration and efficient operation of their facility. This particular document will enable them to capture as much as possible information about the state-of-the-art technology for energy integration known as Pinch Technology.
Page 3 of 208
2.0
2.1
Pinch Technology
History and Background
Pinch Technology/Analysis is developed in the late 1970s as a technique for optimization of thermal heat recovery, and rapidly gained wide acceptance as a theoretically elegant yet practical approach to the design of Heat Exchanger Networks (HENs). Since then, it has evolved into a general methodology for optimization, based on the principles of process integration. It has been applied successfully not only to energy systems (heat recovery, pressure drop recovery, power generation), but also to fresh water conservation, wastewater minimization, production capacity de-bottlenecking, and management of chemical species in complex processes. Applying Pinch Analysis to HEN synthesis and retrofit, engineers can calculate the energy requirement for any process, and produce thermally efficient and practical designs. Energy savings are typically 20% or more compared to previous best designs. Pinch Analysis also applies to optimization of the supply-side, consisting of on-site utilities, such as boilers, furnaces, steam and gas turbines, cogeneration, heat pumps, and refrigeration systems. Consider the example of a process with only two streams as per graph below, one cold (to be heated) and one hot (to be cooled), as illustrated in the figure, and represented on a Temperature-Enthalpy (T-H) diagram. The temperature axis shows the available driving forces for heat transfer, while the enthalpy axis shows the demand for and availability of heat. In this case, both hot and cold duties are supplied by utilities (e.g., steam and cooling water). Now consider recovering some heat from the hot stream to the cold stream. The optimum value of the Minimum Approch Temperature (Tmin) is first determined based on the economic tradeoff between cost savings from heat recovery and capital cost of the heat exchangers. The T-H curves are then moved horizontally until the closest vertical approach between the hot and the cold curves is equal to the Tmin. This point is called the process pinch. The enthalpy overlap represents the optimum amount of heat recovery between the two streams. The residual duties of the two streams must be supplied by utilities, and represent the energy targets. Notice that the duties on both hot and cold utilities are reduced by an identical amount, and equal to the amount recovered. A lower value of Tmin will reduce utility consumption, but will require more heat exchanger area.
Page 4 of 208
For processes with multiple cold streams, the individual process heating duties can be combined into a single cold composite curve drawn on a T-H diagram, which represents the enthalpy demand profile of the process as shown in graph below. Similarly, all the cooling duties (for hot streams) can be combined into a single hot composite curve, which represents the enthalpy availability profile of the process. Composite curves are produced by summing enthalpy changes of individual streams in their respective temperature intervals. When both curves are plotted on the same T-H diagram, they show the opportunity for heat recovery as well as the net heating and cooling targets. The concept of composite curves reduces the multi-stream problem to a 2-stream problem, which we know how to solve. The process pinch separates the overall process into two distinct thermal domains: (a) a net heat sink above the pinch temperature, to which hot utility must be supplied, and (b) a net heat source below the pinch temperature, to which cooling must be provided. In order to achieve the targets, the HEN design must satisfy three conditions: 1) No hot utilities used below the pinch temperature 2) No cold utilities used above the pinch temperature 3) No heat transfer from hot streams above the pinch to cold streams below the pinch
Feed
PROCESS
Product
120
HOT UTILITY
100
80
60
40
20
0 0
COLD UTILITY
10
20
30
40
50
60
Page 5 of 208
Feed
PROCESS
Product
120
HOT UTILITY
100
80
HEAT RECOVERY
60
Pinch
(MAT)
40
20
0 0
COLD UTILITY
10
20
30
40
50
60
The general pinch approach is shown in figure below. The problem is first transposed into the pinch format, which plots composite curves of resource (energy, water, etc.) demand and availability. Targets are set (e.g., for hot and cold utilities consumption), and a broad set of heuristic design rules used to develop a practical design that meets these targets as closely as economically possible. The transposed environment gives the engineer a graphic visualization of even the most complex problems and enables quick assessment of alternatives, including outline economics.
Page 6 of 208
Finally, the pinch environment is transposed back into the familiar flowsheet format, and the final stages of simulation, feasibility checking and detailed design are completed. Pinch Principle in Brief The point where DTmin is observed is known as the Pinch and recognizing its implications allows energy targets to be realized in practice. Once the pinch has been identified, it is possible to consider the process as two separate systems, one system above the pinch and one below it. The system above the pinch requires a heat input from external utilities and is therefore a net heat sink. Below the pinch, the system rejects heat and so is a net heat source. The understanding of the pinch principle gives three rules that must be obeyed in order to achieve the minimum energy targets for a process as we mentioned before, Heat must not be transferred across the pinch There must be no external cooling above the pinch There must be no external heating below the pinch
Violating these soft constraints will lead to cross-pinch heat transfer resulting in an increase in energy requirement beyond the target. These rules form the basis for the pinch design method (PDM) for HEN design. The PDM ensures that there is no cross pinch heat transfer. In retrofit applications the design procedures modify the exchangers that are passing the heat across the pinch. The energy required for the process is supplied through several utility levels. The objective of wise utility utilization in process plants is to maximize the use of the cheapest utility and minimize the use of the expensive one. For instance it is preferable to use low pressure steam rather than high pressure steam and cooling water in lieu of refrigeration. The composite curves introduced before provide overall energy targets but do not clearly indicate how much energy needs to be supplied by different utility levels. Therefore a new tool called the grand composite curve is used for this purpose as will be shown later on.
Page 7 of 208
Net Heat Sink Above the Pinch
Opportunity for heat recovery Net Heat Source Below the Pinch
Basic Concepts for Heat Recovery (New Design) It feels natural to start with the single most important industrial application area for Pinch technology. The development that followed the discovery of the Heat Recovery Pinch has been unique in Process Design when it comes to real life applications in industry based on results from academic research. One of the important advantages of basic Pinch Analysis is that a number of concepts, representations and graphical diagrams have been developed that both are excellent learning aids but also provides the engineer with powerful tools for industrial applications. In order to introduce and explain some of these concepts, representations and graphs, simple examples will be used throughout most of this document.
Page 8 of 208
210 K
160 K
Compressor
130 K
Condenser
270 K
160 K
Reactor
210 K
Reboiler
Feed
50 K
220 K
60 K
Product
The process example shown above illustrates the four main phases of Pinch technology in the design of heat recovery systems for both new and existing processes: 1) Data Extraction, which involves collecting data for the process and the utility system. 2) Targeting for best performance in various respects. 3) Design, where an initial Heat Exchanger Network is established. 4) Optimization, where the initial design is simplified and improved economically. It is important to make sure that a proper problem definition has been formulated This also includes relevant cost data and economic criteria.
Page 9 of 208
The most time consuming and often most critical phase is the identification of the need for heating, cooling, boiling and condensation in the process. This task is more art than science, and if not carried out properly, the final design will not be the best possible. It is quite easy to accept too many features of the proposed flowsheet and blind yourself with a lot of details. This is inevitably will results in the situation where many good opportunities are excluded from the analysis. Once the Data Extraction and corresponding Targeting (Phase 2) activities are completed, it is time to look back and question some of the decisions made for the Reactor and Separation Systems. The idea is then to identify process modifications that will increase the potential for heat recovery and/or allow the use of cheaper utilities. In practice, there are a number of situations where heat integration is not desirable. Examples include long distances (costly piping), safety (heat exchange between hydrocarbon streams and oxygen rich streams), product purity (potential leakage in heat exchangers) switchability, flexibility, controllability, operability and availability. A reasonable strategy is, however, to start by including all process streams and keep the degrees of freedom open. Later, practical considerations can be used to exclude some of these streams and degrees of freedom, and the engineer will then at any time be able to establish the consequences with respect to energy consumption and total annual cost. A central part of data extraction is the identification of heating and cooling requirements in the process. The necessary data for each process stream are the following: m = mass flowrate (kg/s, tons/h, etc.) Cp = specific heat capacity (kJ/kgC) Ts = supply temperature (C) Tt = target temperature (C) Hvap = heat of vaporization for streams with a phase change (kJ/kg) Table below shows the data extracted for the simple example in figure above including data for available utilities, where Q values are variables that will be optimized during targeting and design. In order to analyze area and investment cost for heat exchangers, heat transfer conditions must be established. This is typically done by assigning a film heat transfer coefficient (h) to each process stream. The total coefficient (U) for heat transfer between a hot stream (Hi) and a cold stream (Cj) is then estimated by the simple equation: 1/U = (1/hHi) + (1/hCj)
Page 10 of 208
Stream ID Reactor Outlet H1 Product H2 Feed C1 Recycle C2 Reboiler C3 Condenser H3 HPS MPS LPS Cooling water CW
Ts(C) 270 220 50 160 220 130 250 200 150 15
Tt(C) 160 60 210 210 220 130 250 200 150 20
mCp(kW/C) 18 22 20 50
Q(kW) 1980 3520 3200 2500 2000 2000 (var.) (var.) (var.) (var.)
h(kW/m C) 0.5 0.5 0.5 0.5 1.0 1.0 2.5 2.5 2.5 1.0
An important feature of Process Integration is the ability to identify Performance Targets before the design phase is started. For heat recovery systems with a specified value for the minimum allowable approach temperature (Tmin), targets can be established for Minimum Energy Consumption (external heating and cooling), Fewest Number of Units (process/process heat exchangers, heaters and coolers) and minimum total heat transfer area. In addition, the corresponding calculations will also identify the Heat Recovery Pinch, which acts as a bottleneck for heat recovery. For new designs, it is possible to return to data extraction and modify the process in such a way that the impact of the heat recovery pinch is reduced or even eliminated. Then a new Pinch point will be identified, and the procedure can be repeated. It is also possible to combine targets for energy, units and total heat transfer area into an estimate of the total annual cost. By repeating these calculations for different values of Tmin, it is possible to identify a good starting value for the level of heat recovery. This exercise of pre-optimization has been referred to as "Super-Targeting" (which also gave name to one of the commercial software packages available). While initial methods used a global value for Tmin, later methods allowed individual stream contributions to the overall minimum approach temperature (Ti) reflecting the heat transfer conditions for each process stream, as indicated by its film heat transfer coefficient (hi). One model that has been used is Ti = C / sqrt(hi), where C is a common adjustable factor, reflecting the chosen level of heat recovery. By adding enthalpy changes for the hot and cold process streams in table separately and for each temperature interval in the process, the hot and cold Composite Curves in the graph below can be established. The distillation column (H3, C3) is not included at this stage. How to draw these curves will be illustrated in a step-by-step manner later in the document.
Page 11 of 208
T (C)
Qh-min
Q-recovery
300 250 200 150 100 50 Pinch
Qc-min
0.0 2000 4000 6000
Q (kW)
Composite Curves provide valuable information about maximum heat recovery (QRecovery), minimum external heating (QH,min), minimum external cooling (QC,min) and location of the heat recovery Pinch for a given value of Tmin. Composite Curves can be applied and provide valuable information whenever an amount (such as heat) has a quality (such as temperature). The advantages of graphical representations include a clear aspect of understanding, they provide the engineer with an overview of the problem, they illustrate important economic tradeoffs, and finally they represent information in a very concentrated form. The results (targets) that can be extracted from this graph where Tmin = 20C, are the following: Maximum Heat Recovery: QRecovery = 4700 kW Minimum External Heating: QH,min = 1000 kW Minimum External Cooling: QC,min = 800 kW Pinch Point (caused by a cold stream): TPinch,C = 160C Corresponding Pinch for hot streams: TPinch,H = 180C
Page 12 of 208
As indicated in the previously shown table, the values for mCp are assumed to be constant. This simplifies the calculations from numerical integration to a summation over intervals. When the value of Cp varies considerably with temperature, introducing stream segments can piece-wise Linearise the temperature/enthalpy relation for the stream. The same applies for a stream that has a phase change. Based on the Composite Curves in figure above, a general strategy for Process Modifications can be established. In Pinch Analysis, this strategy has been referred to as the Plus/Minus principle (Linnhoff and Vredeveld, 1984), which means to increase ("plus") heat available above Pinch and/or heat demand below Pinch or to reduce ("minus") heat demand above Pinch and/or heat available below Pinch (see figure below) Examples of such Process Modifications include changes in pressure for distillation columns and evaporators, changes in flowrates for some streams, and new supply and target temperatures for streams when possible. The Plus/Minus principle applied to Composite Curves
QH-min
+ Qc-min
+
Q
Page 13 of 208
While graphical diagrams such as the Composite Curves are excellent tools for learning the methods and understanding the overall energy situation, minimum energy consumption and the heat recovery pinch are more often obtained by numerical procedures. Typically, these are based on the Heat Cascade in graph below, the Heat Cascade, the supply and target temperatures of all process streams divide the temperature scale into Temperature Intervals, in the same way as the construction of the Composite Curves. On the left side of the diagram in graph, hot streams supply heat into the various intervals according to a hot temperature scale. Similarly, on the right hand side of the diagram, cold streams extract heat from the various intervals according to a cold temperature scale. The difference between the hot and the cold temperature scale is the value of the minimum approach temperature, Tmin, thus the heat cascade ensures feasible heat transfer according to an economic criterion.
STEAM H1 720 kW 180 kW 720 kW 360 kW
270C250C
+720
230C210C
500 kW 200 kW 2000 kW 800 kW 400 kW C2 1800 kW
-520
220C200C
C1
880 kW
-1200
180C160C
440 kW 1980 kW
+400
160C140C
+180
70C50C
220 kW
+220
60C40C
CW
H2
The Heat Cascade for part of the process example discussed above capture the essence of the process sources and sinks as the composite diagram does. The objective of this cascade is to allow heat surplus in one interval to cascade down to the next interval, in order to maximize heat recovery. In the graph, a surplus of 720 kW in the first interval can be used to cover the deficit of 520 kW in the second interval.
Page 14 of 208
There is an accumulated heat surplus of 200 kW that can be cascaded further into the third interval. The heat deficit of 1200 kW in the third interval can then be covered by 200 kW of cascaded heat and the inevitable supply of 1000 kW from hot utility, such as steam (ST). The last three intervals in figure above all have heat surplus, and the total heat that must be removed from the cascade in this lower part by cold utility, such as cooling water (CW), is 800 kW. Some important results can be extracted from the Heat Cascade in the subsequent discussion. First, we have identified the need of 1000 kW from hot utility (which, of course, is the same as indicated by the Composite Curves) and 800 kW removed by cold utility. By having this minimum exchange of heat between the process and the utility system, there is no heat flow between intervals 3 and 4. This is the Pinch Point (bottleneck for heat recovery). Above the Pinch temperature (180C/160C), we have a sub-system with heat deficit, and below Pinch there is a subsystem with heat surplus. This decomposition effect is a very important property of the Process Pinch, and it has several important impacts on the design of energy efficient processes. In heat recovery, it becomes important not to transfer heat across pinch. Each single kW taken from the subsystem above Pinch (heat deficit) and transferred to the sub-system below pinch (heat surplus) will immediately require an extra 1 kW of both steam and cooling water. The easiest way to obey this rule is to design two separate heat exchanger networks, one above and one below the Process Pinch. Violations of the Pinch decomposition are also the key to identifying good retrofit projects as will be shown later in details.
Process Cascade Distillation Column Heat Pump Steam Turbine
Q H, min
Q-reboiler
Above Pinch
Q HP, out WHP
Q steam, in
Q = 0.0
W steam
below Pinch
Q HP, in
Q-condenser
Q steam, out
Q c, min
Page 15 of 208
The pinch decomposition characteristics exhibited before can be used to study energy systems with several "Suppliers" and "Customers". The decomposition property has a major impact on the use of Heat Pumps, the use of back-pressure Steam Turbines and the integration of special equipment such as Distillation Columns and Evaporators. The general rule that can be formulated based on the decomposition principle is to try to match heat "suppliers" with heat "customers". Otherwise, heat integration does not serve any energy saving purpose, and will only introduce additional investment cost and less operable processes. This scenario with suppliers and customers is indicated in the shown graph above without any temperature details, however, for each item (process cascade, distillation column, heat pump and steam turbine), temperature is decreasing as we move from the top to the bottom of the graph. Of course, the "supplier" must provide heat at a sufficiently high temperature to meet the needs of the "customer". The following explicit rules are derived from the decomposition principle using graph above. A distillation column should only be integrated with the background process if: a) The reboiler temperature is lower than the Pinch temperature b) The condenser temperature is higher than the Pinch temperature A heat pump should only be integrated with the background process if it takes heat from below Pinch and lifts it above Pinch. Similarly, it should only be integrated with a distillation column if the column can not be integrated with the background process, since heat pumping is more expensive with respect to investment than direct process integration using heat exchangers. A steam turbine should only be integrated (i.e. back pressure or extraction turbine) with a process or distillation column if the outlet steam has a high enough condensing temperature (high enough pressure) to be used above the process Pinch or in a column reboiler. Otherwise a condensing turbine should be used. While the Heat Cascade provides crucial insight about efficient use of energy through heat integration, it is also the basis for an important school of methods based on mathematical models. The heat cascade is a special case of the Transshipment Model which is frequently used in Operations Research and forms the basis for some of the optimization based methods such as Mathematical Programming.
Page 16 of 208
Grand Composite Curve (GCC) Without the Distillation Column STEAM
T (C) 300
T0= 270C Q h, min= 1000
+720
T1= 220C T2= 210C
Thermal Cascade Diagram
R 1= 580 R 2= 580
200
-520 -1200
T3= 170C R 3= 580
Q h, min= 1000
+400
T4=150C R 4= 580
+180
T5=60C R 5= 580
100
+220
T6=50C Q c, min= 800
CW
0.0
Q c, min= 800
500 1000 1500 Q (kW)
The previously discussed decomposition principle has one major disadvantage. We need a better view of the amount of heat available at various temperatures from the "suppliers", and the corresponding need for heat at various temperatures among the "customers". This additional information about Load and Level is crucial for applying the explicit rules mentioned above. It does not help if the distillation column can provide heat above the Process Pinch, if the amount of heat needed at that specific temperature is very small. Such information is implicitly available in the heat cascade however, a much better overview of the situation is obtained if the Heat Cascade is transformed into another graphical diagram called the Grand Composite Curve. This diagram, which also has been referred to as the Heat Surplus Diagram, is generated by plotting so-called modified interval temperatures against the corresponding flow of heat between intervals in the cascade. This is shown in the graph for the example process, after the addition of minimum hot and cold utility requirements. The modified temperatures are simply the average between the hot and the cold temperatures (+/-Tmin /2), an adjustment that allows the drawing of hot and cold streams and utilities in the same temperature scale, while satisfying the need for minimum driving forces.
Page 17 of 208
The Grand Composite Curve has a number of industrial applications, mostly related to the utility system and heat and power considerations. Typically, the Grand Composite Curve can be used to qualitatively and to some extent quantitatively address the following tasks: Identify a near-optimal set of utility types (both load and level) to cover the need for external heating and cooling in the process. A Utility Grand Composite Curve consisting of available utilities, such as for example various steam levels, flue gas from a furnace or gas turbine, hot oil circuits, cooling water, refrigeration, etc., can be combined in such a way that total utility cost is minimized. Identify potential for steam production below Pinch, if the process Pinch is at a Sufficiently high temperature. This means that steam generation (typically LP steam) is acting as a cold utility. Identify potential for utilizing so-called "pockets" in the Grand Composite Curve for additional power production. If the temperature difference had been sufficiently large between the part of the process where there is local heat surplus and the corresponding part where there is local heat deficit, there would have been some scope for producing steam that could have been used in a back pressure turbine. The turbine then borrows steam generated in the process and returns steam for heating at a lower level after power production. Identify scope for using heat pumps in the process to reduce both hot and cold utility consumption. Typically, this is the case where there is a distinct Pinch point, with flat profiles both immediately above and below the Pinch. In such cases, a significant amount of heat can be transferred from the heat surplus region below Pinch to the heat deficit region above Pinch, by using a heat pump with moderate temperature lift. Identify whether there is scope for integration of special equipment such as distillation columns or evaporators with the background process.
Page 18 of 208
Grand Composite Curve with box representation for distillation columns is shown below.
Returning to the process example, the Grand Composite Curve can be used to give a quick and simple answer about the scope for integrating the distillation column with the background process, or whether it should be operated with utilities (steam and cooling water). After heat integration with the process has been analyzed, the next step could be to evaluate the scope for heat pumping. The graph above shows the Process Grand Composite Curve and the Temperature/Enthalpy Diagram for the distillation column in our process example. Since the distillation column operates across the Pinch, there will be no energy savings from integration with the process. This also follows from the decomposition concept illustrated before. The graphical representation in figure above has also been referred to as the Andrecovich diagram which is a later extension within Pinch Analysis include a refinement of the box representation, where a Column Grand Composite Curve (CGCC) shows the need for re-boiling and condensation at various temperatures in the column. The CGCC is based on converged profiles from a rigorous column simulation,
Page 19 of 208
and can be used to identify the scope for distributed reboiling and condensing as well as feed pre-heating or pre-cooling, and finally changes in the reflux ratio for the column. Process and Utility Grand Composite Curve for the Process Example
Referring to the simple process example and the list of available utilities in the table, the Grand Composite Curve shown above can, as explained, be used to identify the set of utilities with minimum energy cost. Typically, this means to maximize the use of cheaper utilities in order to minimize the use of more expensive utilities. This is shown in graph below where the amount of MP steam is maximized and limited by the situation where the MP part of the Utility Grand Composite Curve touches the Process Grand Composite Curve. Similarly, a potential for LP steam production is identified below Pinch, and the amount is again limited by the point where the LP part of the Utility Grand Composite Curve touches the Process Grand Composite Curve.
Page 20 of 208
If we assume that utility prices are 200 $/kWyr for HP steam, 170 $/kWyr for MP steam, 140 $/kWyr for LP steam and 20 $/kWyr for cooling water, the annual energy cost for the utility mix in figure 5.8 is 166,000 $/yr. When using HP steam and cooling water only, the corresponding annual energy cost is 216,000 $/yr, i.e., 30% higher. The Grand Composite Curve enables the engineer to identify a set of utilities that gives minimum energy cost. As always, however, there is a trade-off between operating cost (energy) and investment cost (number of heat exchangers and their total heat transfer area).Thus, the following important factors need further investigation before accepting the set of utilities proposed in graph: Temperature driving forces will be reduced when introducing MP and LP steam, which means larger heat transfer area in some utility and process/process exchangers. As a result, there will be a significant increase in the investment cost. New Utility Pinch Points will be introduced when maximizing MP steam usage and LP steam production. This will result in tighter designs and more complex heat exchanger network structures. The decomposition feature of the Process Pinch also applies to Utility Pinches. This means for example that heat pumps can be used to transfer heat across (from below to above) all Pinch points in order to reduce total heating and cooling requirements (Process Pinch) or reduce the need for a more expensive utility (Utility Pinch). The number of heat transfer units will increase whenever new utilities are introduced and whenever Utility Pinches are created, which means increased investment cost. The complexity of the heat exchanger network (number of units, piping and stream splits) will increase with an increasing number of Pinch points included during design. While significant savings in energy cost can be obtained by introducing intermediate (and thus cheaper) utilities, there will be a corresponding increase in investment cost (total heat transfer area and the number of units will increase). Minimum total annual cost is found by exploring these trade-offs. The Grand Composite Curve (GCC) has the inherent limitation (which also in many respects is an advantage) that details about the individual streams are not shown. Thus, any conclusion about integration of distillation columns and heat pumps as well as steam generation, must be evaluated carefully by looking beyond the GCC and into the actual number of streams that would be involved. If a heat pump would have to extract (deliver) heat from (to) a large number of streams, it would not be economically interesting. The same applies if we end up with a large number of steam boilers.
Page 21 of 208
2.2
Data Extraction
The objective of this part is to show in details how to extract the right data to obtain energy targets for the hot and cold utilities consumption and select the optimal utility scheme for a desired plant or process area. This can be essentially done via three methods, graphical, algebraic and using mathematical programming/optimization. Step number zero in our calculation is to extract from the process flow diagrams and other process information data sheets the necessary data to conduct our targeting step. This step is called in Pinch technology data extraction. There are two main categories of data extraction. The first set involves the extraction of the flow rate-specific heat and supply and target temperatures for a given process to conduct heat and power integration study. The second set involves the extraction of process conditions that challenges every step in the process to design/retrofit the scheme for the sake of energy efficiency optimization. In general, data extraction involves collecting data for the process and the utility system. It relates to the extraction of information required for pinch analysis from given process heat and material balance. This phase is most time consuming phase and it is very important for the identification of process needs from heating, cooling, condensation and evaporation utilities. The extraction of the necessary thermal data from any flow sheet includes the identification of the cooling and heating utilities required. In such case the flow sheet will be drawn in a way that highlights the cooling and heating utilities demands without any reference to cooler, re-boilers and heat exchangers. _Each process stream needs to be defined either as a hot or cold stream. Such definition depends upon the task needed to be conducted regarding the stream. If it is raising the stream temperature then the stream will be considered as a cold stream. Otherwise (to reduce its temperature) the stream is going to be considered a hot stream.
Page 22 of 208
If the task is condensing a vapor stream or mixed (vapor and liquid) then it is a hot stream. If the task is vaporizing a liquid stream or mixed (vapor and liquid) then the stream is a cold stream. _Upon defining the type of each stream (hot or cold) and create a list of hot streams and cold streams; the flowrate of each stream (F) need to be extracted from the process flow diagrams. Note: DCS real-time graphics can be used to extract the mass flowrate of the desired streams. _ Thermal properties of the streams such as specific heats (Cp) and latent heats of vaporization and condensation need also to be obtained from thermodynamic packages or literatures. _ It is also important to consider stream composition to be able to get accurate thermal properties for desired process streams. _ Streams supply temperatures (Ts) and target temperatures (Tt) need to be defined for each hot and cold stream These thermal data are depicted in the table below: Strm.# Strm.Type Start Temp(Ts)(C) Target Temp.(Tt)(C) Heat Capacity Flowrate(kW/C) 1 hot 180 80 20 2 hot 130 40 40 3 cold 60 100 80 4 cold 30 120 36 DT min= 10C, Utility steam is supplied at 200C and Cooling water at 25C to 30C This table shows the data necessary to obtain energy targets.
Page 23 of 208
Hot streams are the streams that need to be cooled (heat sources). Cold streams are streams that need to be heated (heat sinks). The heat capacity flow rate (FCp) is the mass flow rate times the specific heat capacity. Where (Cp) is the stream specific heat capacity (kJ/C, kg) and (F) is the mass flow rate (kg/sec). The (FCp) of a stream is measured as enthalpy change per unit temperature (kW/C). The following process example will be used as an illustration to the data extraction step from the process flow diagram shown below. Remember for each stream we need to define the mass flow rate (kg/sec), supply temperature (C), target temperature (C), specific heat capacity (kJ/kgC) and heat of vaporization or condensation for streams with phase change (kJ/kg). Stream/Type Start Temp(Ts)( C) Target Temp.(Tt)( C) Heat Capacity Flowrate(kW/ C) Reactor outlet (hot) 180 80 20 Product (hot) 130 40 40 Feed (cold) 60 100 80 Recycle (cold) 30 120 36 These thermal data are mandatory to conduct energy targeting study. This data extraction should be done carefully since poor data extraction can results in missed opportunity for improved process design for energy saving.
Page 24 of 208
210 K
160 K
Compressor
130 K
Condenser
270 K
160 K
Reactor
210 K
Reboiler
Feed
50 K
220 K
60 K
Product
Over the years pinch technology experts developed some heuristics for data extraction. Some of these heuristics are as follows: 1- Do not carry over features of the existing solution 2-Linearise on the safe side 3-Do not extract utilities 4-Do not mix streams at different temperatures. 5-Define soft data 6-Use targeting to determine cost of constraints 7-Extract at effective temperature
Page 25 of 208
Let us explain these heuristics one-by-one and with some details. 1- Extract streams from flow-sheet according to deep understanding of the process and not based upon the existing Heat exchangers network 2- The enthalpy change of some streams is significantly non-linear. This is particularly true for condensing/vaporizing streams. In such situations, adopting just one value of FCp might lead to inaccurate results. It is more accurate to represent the stream in as many segments as is required to closely mimic the heating and cooling curve of the stream. These segments should exhibit the following rules to guarantee safe linearization. The actual hot stream must be hotter than the extracted hot stream The actual cold stream must be colder than the extracted stream
Note: Phase Changes: The pinch procedures require that the FCp values of streams be constant. We can incorporate phase changes that take place at constant temperature into this formulation simply via assuming a 1F temperature change at the temperature of the phase change and then calculating a fictitious FCp value that gives the same heat duty as the phase change. In other words if the heat corresponding to the phase change is FHv we write FfCpf(1)= FHv; Where Ff and Cpf are the fictitious values. For the case of mixtures, where a plot of enthalpy versus temperature is curved, we merely linearize the graph and select fictitious FCp values that have the same heat duty.
Page 26 of 208
Original hot streams are hotter than proposed hot segments
Original Cold streams are colder than proposed cold segments
3- A true utility stream is a utility stream like steam, cooling water, refrigerant, brine,etc., that can be in principle replaced by any other stream(utility or process) for heat exchange purposes. Such true utility streams should not be extracted as a part of the process streams that need to be cooled or heated. An example of true utility stream is cooling water used in a heat exchanger. Such stream can be replaced by any other process stream heating or air cooling, refrigerant cooling. This stream should not be extracted. In the other hand when steam is used as a raw material in methane- steam reforming process the steam is not a true utility stream that can be replaced by other utility stream. In such situations the steam must be extracted as a process cold stream that need to be heated from boiler feed water conditions to the appropriate temperature for the reformer/reactor. However, for the sake of overall steam consumption reduction, the process should also be challenged from design and/or operational point of view to look for more efficient energy-conscious alternatives. (i.e., dry reforming)
Page 27 of 208
Another example is when direct live steam is used for stripping. If this steam is used for just heating purposes for reboiling a column and can be replaced by heating oil, other heating medium or process cooling stream it would be treated as true utility stream. If the reboiling must be via direct injection of steam, then the steam is not a utility stream and should be extracted as part of the process cold streams that need to be heated from water to the desired target temperature. 4- One common problem in data extraction occurs when the process flow-sheet involves the mixing of process streams at different temperatures. When the streams are at different temperatures, the mixing junction then acts as direct contact heat exchanger. Thermodynamically it is not recommended to mix streams at very different temperatures. First because of the high level of entropy generation resulting in losses in availability widely known as exergy or lost work. Second the mixing junction could involve cross-pinch heat transfer and therefore an increase in the overall process energy requirements as we mentioned before. To avoid this situation it is recommended that if mixing has to happen any way due to some unchallengeable process reason, the streams involved in such mixing process should be extracted as being mixed isothermally. In other words these streams need to reach the same temperature before get mixed. The figure below illustrates this rule.
Page 28 of 208
NO
20 220 110 20 60
YES
20
110 110
NO
200
20
110
150 100 200 T
50
Look for the temperature that is beneficial To the energy targets
50 100 T
5- The temperature, pressure and enthalpy conditions of some streams within a process are open to change within certain limits. Such streams data are called soft data. An example of soft data is the pump discharge pressure upstream (immediately) of a vaporizer. The pump discharge pressure can be varied within certain limits resulting in a corresponding flexibility in the vaporization temperature. Another example of soft data is the target temperature of a product stream which is to go into storage. This temperature can be within a substantial temperature range. These soft data should ideally be extracted such that the overall process energy requirement is minimized.
Page 29 of 208
6- Process constraints can have a detrimental effect on process energy requirement. Energy targeting tools can be used to investigate of certain constraint on the process energy demand. For instance in some cases it is preferred to keep process areas independent of each others. In such cases the pinch targeting tool can be utilized to explore the effect of such constraint on the process energy requirement. The example below demonstrates such calculation.
What is the cost of separating A from B
A+B
Energy Consumption= 200 HEN Area= 120
Energy Consumption=160 HEN Area= 132
Page 30 of 208
7- For data extraction purpose the effective stream temperatures are more important than the actual stream temperatures. For example, for a hot stream it is important to know what temperature it is available at to exchange heat against cold streams, rather than its actual temperature. Similarly, for cold streams it is the temperature at which heat must be supplied to them that is important. A famous example if you have a reactor outlet stream (product) that must be quenched during which steam is raised. As a result of this constraint the temperature at which the heat in the product stream is available to other streams is not the reactor outlet temperature but the temperature at which steam is raised. Therefore, extracting this process stream at the effective temperature which is the condensing steam temperature will gives the correct energy targets. In other words due to given process constraint the product stream would have a very large stream DTmin with the cooling/quenching water stream. Notes on Extracting Data from PFDs: *First Category: _Exchangers which operate in parallel can be considered as a single stream with an enthalpy change equal to the combined duties of the existing exchangers and a temperature profile from the initial stream temperature to the final mixed temperature. _Exchangers operating in series can also be combined into single stream, starting at the coldest and ending at the hottest. Duties of the individual exchangers may be considered as segments, if the stream thermal profile is not linear. Upon the completion with the data extraction phase now we can start our targeting phase using Pinch technology. 2.3 Targeting
Graphical Method As we explained before any heat exchanger can be represented as a hot stream that is cooled down by another cold stream and/or cold utility and a cold stream that is heated up by a hot stream and/or hot utility with a specified minimum temperature approach between the hot and the cold called Tmin. The process exhibited below in the graph shows the situation when the two streams do not have a chance of overlap that produce heat integration between the hot and the cold.
Page 31 of 208
Feed
PROCESS
Product
120
HOT UTILITY
100
80
60
40
20
0 0
COLD UTILITY
10
20
30
40
50
60
Moving the cold stream to the left on the enthalpy axis without changing its supply and target temperatures till we have small vertical distance between the hot stream and the cold stream we obtain some overlap between the two streams that result in heat integration between the hot and the cold and less hot and cold utilities. As been depicted in the graph below with shrinkage in the red and blue lines span.
Page 32 of 208
Feed
PROCESS
Product
120
HOT UTILITY
100
80
HEAT RECOVERY
60
Pinch
(MAT)
40
20
0 0
COLD UTILITY
10
20
30
40
50
60
Now we want to represent all the hot streams in the process by one long hot stream and we will call this line the hot composite curve. We will also do the same thing with all the cold streams in the process. The next step will be drawing the two composite curves/lines on the same page in Temperature (T)-Enthalpy diagram with two conditions: 1- The cold composite curve should be completely blow the hot composite curve, and 2- The vertical distance between the two lines/curves in terms of temperature should be greater than or equal to a selected minimum approach temperature called global Tmin
Page 33 of 208
The resulting graph is depicted below and known as thermal pinch diagram
Net Heat Sink Above the Pinch
Opportunity for heat recovery Net Heat Source Below the Pinch
This step can be done in details as follows: Constructing the composite curves (step-by-step) Stream Type Supply Temperature (C) Target Temperature (C) 1-Hot 2-Hot 3-Cold 4-Cold 170 120 50 20 70 30 90 110 FCp (kW/C) 10 20 40 18
Page 34 of 208
1- Draw the hot composite curve and the cold composite curve via developing the following tables. Note: The tables list all the hot and cold streams temperatures in an ascending order with the cumulative enthalpy corresponding to the lowest hot temperature and lowest cold temperature respectively equal to zero. 2- In every temperature interval the cumulative hot load is calculated using the following formula H= FCp * (Tsupply Ttarget) 3- In every temperature interval the cumulative cold load is calculated using the following formula H= FCp * (Ttarget Tsupply) Hot streams temperature list T0=30 T1=70 T2=120 T3=170 Cold streams temperature list T0=20 T1=50 T2=90 T3=110 Cumulative Enthalpy (H) H0=0.0 H1=800 H2=2300 H3=2800 Cumulative Enthalpy (H) H0=0.0 H1=540 H2=2860 H3=3220
Page 35 of 208
Temperature (T)- Enthalpy (H) Diagram

T
Hot composite curve
Cold composite curve
30 20 Cold composite curve is not completely below the hot composite curve
As we mentioned before the cold composite curve shall lie completely below the hot composite curve and this can be done via dragging the cold composite curve to the right on the enthalpy axis (H). This process shall stop at a vertical distance between the cold and the hot composite curve for a temperature equal to the minimum temperature approach selected earlier.
Page 36 of 208
Temperature (T)- Enthalpy (H) Diagram

Minimum Heating Utility T Qh=480 kW
Hot composite curve
Cold composite curve
Minimum Temperature Approach 30 20 Qc=60 kW Minimum Cooling Utility H
Algebraic Method: Information needed Given a unit with a list of hot streams to be cooled and cold streams to be heated Stream ID Type (FCp) 1 2 3 4 hot hot cold cold 10 5 19 2 Supp. Temp. Target Temp. 520 380 300 320 330 300 550 380
Page 37 of 208
Constructing temperature interval diagram 1.1_ Draw two temperature scales one for the hot streams and another for the cold streams 1.2_ Select reasonable minimum temperature approach between the hot streams and the cold stream (for instance, 10 C) 1.3_ Draw all the hot streams (in the table hot section) to be cooled according to the hot steam scale as arrows that start at the supply temperatures and end at the target temperatures 1.4_ Repeat step 1.3 for all cold streams in the cold section of the table 1.5_ Start at the highest temperature of any hot stream in the hot section and draw a horizontal line that span along the two sections of the table, the hot and the cold. 1.6 Draw horizontal lines again at the start and the end of any arrow representing the hot streams in the hot section of the table 1.7 Repeat step 1.6 for any arrow representing cold stream in the cold section (at the start and the end of any arrow) 1.8 Count the number of segments generated and number them starting at the highest temperature (they are called temperature intervals) 1.9 Make sure that each temperature interval has now temperature value on both the hot temperature scale and cold temperature scale. The difference is the desired minimum temperature approach (for instance the 10C used in this example) These procedures are depicted in the figure below Note: This structure means that within any temperature interval it is thermodynamically feasible to transfer heat from the hot streams to cold streams. It is also feasible to transfer heat from a hot stream in an interval x to any cold stream which lies in an interval below.
Page 38 of 208
The temperature Interval Diagram
T minimum = 10 K
T* 555 515
Interval
Hot Streams 560 H1 520
Cold Streams 550 510
2 385 375 310 305 295 3 4 5 6 H2 390 380 330 310 300 380 370 320 300 290 C1 C2
Hot Streams:H1; F1Cp1= 10 kW/K H2; F2Cp2= 5 kW/K
Cold Streams:C1; F1Cp1= 10 kW/K C2; F2Cp2= 5 kW/K
Note: The temperature symbol T* is interval inlet temperature used later on selecting the suitable energy utility after calculating the targets using what is known as grand composite curve. To calculate T* we take the average interval inlet temperature of the hot and cold temperature scale.
Page 39 of 208
Constructing tables of exchangeable heat loads and cooling capacities 2.1 Determining individual heating loads and cooling capacities of all process streams for all temperature intervals using this formula: Qnm = F1Cp1* (Ts-Te) in energy units ( kW) Ts is the interval start temperature and Te is the interval end temperature n is stream number and m is the interval number Example 1: Interval # 1 in the hot section: The interval start temperature is 560 K The interval end temperature is 520 K Q11(Q for stream #1 in interval #1)= F1Cp1*(560-520) Since there is no H1 stream in this interval, hence, F1Cp1=0.0 Q stream # 1(exchangeable load) in this interval = 0.0*(560-520)= zero Example 2 Interval # 2 in the hot section: The interval start temperature is 520 K The interval end temperature is 390 K The flow specific heat F1Cp1= 10 kW/K Then, Q stream #1(exchangeable load) in interval #1= 10*(520-390) = 1300 kW
Page 40 of 208
Example 3 Interval # 1 in the cold section: The interval start temperature is 550 K The interval end temperature is 520 K The flow specific heat of this cold stream is F1Cp1= 119 kW/K Then, Q stream #1(cooling capacity) in interval #1= 19*(560-520) = 760 kW Upon the completion of this step 2.2 We can now obtain the collective loads (capacities) of the hot (cold) process streams. These collective loads (capacities) are calculated by summing up the individual loads of the hot process streams that pass through that interval and the collective cooling capacity of the cold streams within the same interval
Page 41 of 208
These calculations for the above problem is shown in the following tables
Table of Exchangeable Loads for Process Hot Streams Intervals
Interval 1 2 3 4 5 6
Load of H1, kW 0.0*(560-520)= 0.0
Load of H2, kW 0.0*(560-520)= 0.0
Total Load, kW 0.0+0.0= 0.0 1300+0.0= 1300 100+0.0= 100 500+250= 750 0.0+ 100= 100 0.0+50= 50
10*(520-390)= 1300 0.0*(520-390)= 0.0 10*(390-380)= 100 10*(380-350)= 500 0.0*(330-310)= 0.0 0.0*(310-300)= 0.0 0.0*(390-380)= 0.0 5*(380-330)= 250 5*(330-310)= 100 5*(310-300)= 50
Page 42 of 208
Table of Cooling Capacities for Process Cold Streams Intervals
Interval 1 2 3 4 5 6
Capacity of C1, kW Capacity of C2, kW 19*(550-510)= 760 0.0*(550-510)= 0.0
Total Load, kW 760+0.0= 760 2470+0.0= 2470 190+20= 210 950+100= 1050 380+ 0.0= 380 0.0+0.0= 0.0
19*(510-380)= 2470 0.0*(510-380)= 0.0 19*(380-370)= 190 19*(370-320)=950 19*(320-300)= 380 0.0*(300-290)= 0.0 2*(380-370)= 20 2*(370-320)= 100 0.0*(320-300)= 0.0 0.0*(300-290)= 0.0
Constructing thermal cascade diagrams This diagram is constructed using the total hot loads and cooling capacities obtained in the previous step for each temperature intervals. The temperature intervals are drawn as rectangular with two inlets and two outlets. The inlet from the left is the total hot load available in this interval (for instance, 1300 kW in case of interval # 2) The inlet from above is the utility input load, in case of the first interval, or the input from interval above in case of second, third,,N intervals. The output from the right is the total cooling capacity of this interval (for instance, 2470 kW in case of interval #2).
Page 43 of 208
The output from the bottom is the difference between the total inputs and the cooling capacity of the interval The heat balance around each interval will be conducted as follows
First Interval Heat Balance
Hot Load From Utility Source Top Input
Hot Load From Process Source Left Input
Cooling Capacity From Process Source
Right Output
Residual Hot to Subsequent Interval Bottom Output from first interval
Heat Balance Top Input+ Left Input- Right Output = Bottom Output
Page 44 of 208
Numerical Example of First Interval Heat Balance

Hot Load From Utility Source Top Input= 0.0 kW
Hot Load From Process Source Left Input= 0.0 kW
Right Output=760 kW
Residual Hot to Subsequent Interval Bottom Output from first interval
= - 760 kW
Heat Balance Top Input+ Left Input- Right Output = Bottom Output 0.0 + 0.0 - 760 = - 760 kW
Page 45 of 208
Subsequent Intervals Heat Balance
Hot Load From Above Interval Top Input
Hot Load From Process Source Left Input
Right Output
Residual Hot to Subsequent Interval Bottom Output
Heat Balance Top Input+ Left Input- Right Output = Bottom Output
Page 46 of 208
Numerical Example for Subsequent Intervals Heat Balance For instance; Interval # 2
Hot Load From Above Interval Top Input = -760
Hot Load From Process Source Left Input= 1300
Right Output= 2470
Residual Hot to Subsequent Interval Bottom Output = - 1930
Heat Balance Top Input+ Left Input- Right Output = Bottom Output - 760 + 1300 -2470 = -1930
Page 47 of 208
Upon the completion of heat balance around each interval the following diagram will be produced.
Thermal Cascade Diagram (Un-Balanced) Note: During this step the input from Hot Utility to the first interval is equal to zero 0.0 0.0 1300 100 750 100 50 1 - 760 2 - 1930 3 - 2040 4 -2340 5 - 2620 6 - 2570 0.0 380 1050 760 2470 210
The maximum difference between the available hot loads and cooling capacities from the heat balances of these intervals is 2620 kW. This deficiency in heat will be supplied via outside hot utility.
Page 48 of 208
This value will be the input (from the top of the first interval) and the same heat balance calculation conducted above will be repeated to produce the balanced thermal cascade diagram below.
Thermal Cascade Diagram (Balanced) Note: During this step the input from Hot Utility to the first interval is equal to zero Minimum Q-heating = 2620 kW 0.0 1300 100 750 100 50 760 1860 2 690 3 580 4 280 Thermal Pinch 5 0.0 6 0.0 380 1050 2470 210
Minimum Q-cooling = 50 kW
With the completion of this step, now the minimum heating utility and minimum cooling utility required are 2620 kW and 50 kW respectively.
Page 49 of 208
Constructing the grand composite curve (G.C.C) This curve will be drawn between T* calculated before and the corresponding top heat inputs to each interval. These data are depicted below:
Data Required To Construct The G.C.C
T* (K) 555 T* (K) 515 385 3 375 4 310 305

Thermal Pinch
Enthalpy ( kW) 2620 kW 1 1860 kW 2 690 kW 580 kW 280 kW 5 6 0.0 kW
295
50 kW Enthalpy ( kW)
Page 50 of 208
Drawing these data as T* versus Enthalpy results in the following diagram that can be used to define different levels of utilities that can be used to satisfy the process heating utility requirement as shown below.
Grande Composite Curve (G.C.C)

Should Be Drawn To Scale
T* (K) 600
Total hot utility required is equal to 2620 kW Hu3 Hu2
500 Hu1 400
300
200
Enthalpy ( kW) 700 1400 2100 2800
Page 51 of 208
Multiple utility targeting/selection using Grand Composite Curve (GCC) Upon maximizing heat recovery in the heat exchanger network, those heating duties and cooling duties not serviced by heat recovery must be provided by external utilities. The most common utility is steam. It is usually available at several levels. High temperature heating duties require furnace flue gas or a hot oil circuit. Cold utilities might be refrigeration, cooling water, air cooling, furnace air preheating, boiler feed water preheating, or even steam generation at higher temperatures. Although the composite curves can be used to set energy targets, they are not a suitable tool for the selection of utilities. The grand composite curve drawn above is a more appropriate tool for understanding the interface between the process and the utility system. It is also as will be shown in later chapters a very useful tool in studying of the interaction between heat-integrated reactors, separators and the rest of the process. The GCC is obtained via drawing the problem table cascade as we shown earlier. The graph shown above is a typical GCC. It shows the heat flow through the process against temperature. It should be noted that the temperature plotted here is the shifted temperature T* and not the actual temperature. Hot streams are represented by Tmin/2 colder and the cold streams Tmin/2 hotter tan they are in the streams problem definition. This method means that an allowance of Tmin is already built into the graph between the hot and the cold for both process and utility streams. The point of zero heat flow in the GCC is the pinch point. The open jaws at the top and the bottom represent QHmin and QCmin respectively. The grand composite curve (GCC) provides convenient tool for setting the targets for the multiple utility levels of heating utilities as illustrated above.
Page 52 of 208
The graphs below further illustrate such capability for both heating and cooling utilities.
The above figure (a) shows a situation where HP steam is used for heating and refrigeration is used for cooling the process. In order to reduce utilities cost, intermediate utilities MP steam and cooling water (CW) are introduced. The second graph (b) shows the targets for all the utilities. The target for the MP steam is set via simply drawing a horizontal line at the MP steam temperature level starting from the vertical axis until it touches the GCC. The remaining heat duty required is then satisfied by the HP steam. This maximizes the MP steam consumption prior to the remaining heating duty be fulfilled by the HP steam and therefore minimizes the total utilities cost. Similar logic is followed below the pinch to maximize the use of the cooling water prior the use of the refrigeration.
Page 53 of 208
The points where the MP steam and CW levels touch the GCC are called utility pinches since these are caused by utility levels. The graph (C) below shows a different possibility of utility levels where furnace heating is used instead of HP steam. Considering that furnace heating is more expensive than MP steam, the use of the MP steam is first maximized. In the temperature range above the MP steam level, the heating duty has to be supplied by the furnace flue gas. The flue gas flowrate is set as shown in graph via drawing a sloping line starting from the MP steam to theoretical flame temperature Ttft. If the process pinch temperature is above the flue gas corrosion temperature, the heat available from the flue gas between the MP steam and pinch temperature can be used for process heating. This will reduce the MP steam consumption. In summary the GCC is one of the basic tools used in pinch technology for the selection of appropriate utility levels and for targeting for a given set of multiple utility levels. The targeting involves setting appropriate loads for the various utility levels by maximizing cheaper utility loads and minimizing the loads on expensive utilities.
Page 54 of 208
(C) T*
T-tft
MP
CW
Refrigeration
Normally, we have a choice of many hot and cold utilities and the graph below shows some of our options. Generally it is recommended to use hot utilities at the lowest possible temperature while we generate it at the highest possible temperature. And for the cold utilities it is recommended to use it at the highest possible temperature and generate at the lowest possible temperature. These recommendations are best addressed systematically using the grand composite curve.
Page 55 of 208
Hot and cold utilities

Boiler House And Power Plant Fuel
Steam Turbines
W Gas Turbines W
Hot Oil Circuit
BFW preheat
Process
Heat Pump
Furnace
Cooling Towers Air preheat Refrigeration
Understanding the Grand Composite Curve The graph below shows that utility pinches are formed according to the number of utilities used. Each time a utility is used a utility pinch is created. It also shows that the GCC right noses sometimes known as pockets are areas of heat integration/energy recovery. In other words it does not need any external utilities. These right noses/pockets are caused by; _ Region of net heat availability above the pinch _ Region of net heat requirement below the pinch
Page 56 of 208
Applying the Grand Composite Curve GCC curve can be used by engineers to select the best match between utility profile and process needs profile. For instance, the steam system shown below needs to be integrated with the process demands profile to minimize low pressure steam flaring and high or medium pressures steam let downs. Besides it helps selecting steam header pressure levels and loads.
Page 57 of 208
HP Boiler
HP
Proc. #1
HP Process Condensate
Proc. #2
chemicals
Vent
MP Boiler
MP
Proc. #4 Proc. #1 Vent Deaerator
MP Process Condensate
BFW Raw water Make-up Treatment Plant
LP
Effluent
Process Condensate
Proc. #1
Proc. #3
LP Process Condensate
Page 58 of 208
Superimposed Utility Profile with Process Profile Nominal Case Supply-Demand Matching Problem
T
HP
MP
Process GCC
LP
BFW CW
H
The superimposed steam system on the process grand composite curve shows that while process heating needs can be achieved electricity can also be generated to satisfy process demands and/or export the surplus to the grid. The graph below shows how we can use the GCC not only to select utility type, load but also to define the steam headers minimum pressure/temperature to minimize driving force and save exergy.
Page 59 of 208
Qh
HP MP
LP
BFW
CW
Qc H
Grand Composite Curve can also be utilized to select the load and return temperature of hot oil circuits. The graph below shows that while in many cases the process pinch can be our limiting point in defining the load (slop of the hot oil line) and the return temperature of the heating oil. In some other cases the topology of the GCC is the limiting point not the process pinch. This is also shown in the second graph below. This practical guide to select the load and the target temperature of the hot oil circuits is also applicable to furnaces as will be shown later in this chapter.
Page 60 of 208
Process Pinch temperature is the Limiting temperature for the Hot oil return temperature
T*
T supply Hot Oil
T return Process Pinch
CW
Refrigeration
Process Pinch temperature is not the Limiting temperature for the Hot oil return temperature But the topology of the GCC curve
T*
Hot Oil CP-min T return
T supply
Process Pinch
CW
Refrigeration
Page 61 of 208
Grand composite curve (GCC) can also be used to select the process refrigeration levels and the synthesis of the multiple-cycles refrigeration systems as we did in the steam system. The schematic graph below shows a simplified refrigeration system.
Schematic Diagram for multi-level Refrigeration System
Condenser
25C
CW -5C
Process 0C
Process -35C
-40C
Process -65C
-70C
Work Compressor
Page 62 of 208
The GCC as we mentioned before can be used to place the refrigeration levels as we did with steam levels. The graph below shows how we can do that.
We can place the refrigeration levels like steam levels. Maximizing the highest temperature load to minimize the lower temperature loads
Tcw
- 5 C
- 40 C - 70 C
When a hot utility needs to be at a high temperature and/or provide high heat fluxes, radiant heat transfer is used from combustion of fuel in furnace. Furnace designs vary according to the function of the furnace, heating duty and type of fuel, and method of introducing combustion air. Sometime the function is purely to provide heat; sometimes the furnace is also a reactor like the steam-methane reformer and the furnace provides the heat of reaction.
Page 63 of 208
Process furnaces have a number of common features. In the chamber when the combustion takes place the heat is transferred mainly by radiation to tubes around the walls of the chamber, through which passes the fluid to be heated. After the flue gas leaves the combustion chamber, most furnace designs extract further heat from the flue gas in a convection section before the flue gas is vented to the atmosphere. A simplified furnace model is shown in the graph below.
T0- corrosion
T-stack
T-TFT
Q-Hmin
Air
Fuel
T-TFT is the theoretical flame temperature, while T0-corrosion is the minimum flue gas outlet temperature that still avoids furnace stake corrosion.
Page 64 of 208
Furnace flue gas used as a hot utility can be matched against the grand composite curve as shown below.
T* T*TFT
Flue Gas
T*stack T*ambient Stack loss Fuel Enthalpy Q-Hmin
The flue gas starts at its theoretical flame temperature shifted with the desired Tmin on the grand composite curve and presents a sloping profile because it is giving up sensible heat. Theoretical flame temperature is the temperature attained when a fuel is burnt in air/oxygen without any heat loss. The real flame temperature will be lower than the theoretical flame temperature because heat is lost from the flame because part of the heat released provides heat for a variety of endothermic dissociation reactions that occur at high temperatures. Thus although theoretical flame temperature is not very accurate representation of the true flame temperature, it dose provides a convenient reference to indicate how much heat is actually released by the combustion as the flue gas is cooled.
Page 65 of 208
In figure above the flue gas is cooled to pinch temperature before being released to the atmosphere. The heat released from the flue gas between the pinch temperature and ambient temperature is the stack loss. These losses can be easily calculated for a given process grand composite curve and given theoretical flame temperature. Combustion processes work with an excess air or oxygen to ensure complete combustion of the fuel .Excess air typically ranges between 5 to 20% depending on the fuel, burner design and furnace design. As excess air is reduced the theoretical flame temperature increases. This action has the effect of reducing the stack losses and increasing the thermal efficiency of the furnace for a given process duty. Alternatively, if the combustion air is preheated again the theoretical flame temperature increases and the stack losses decrease resulting in better furnace efficiency as shown in the graph below.
T* T*TFT2
T*TFT1
Flue Gas
T*stack T*ambient Enthalpy Q-Hmin Fuel
Stack loss
Page 66 of 208
In the two furnace-GCC graphs shown above the flue gas is capable to be cooled to the pinch temperature before releasing it to the atmosphere. This is not always the case. One famous case is the situation in which the acid dew point temperature, which is the practical minimum temperature to which the flue gas can be cooled without condensation that causes corrosion in the stack, is higher than the pinch temperature as shown in graph below.
T* T*TFT
Flue Gas
T*ambient
T*stack=T*corrosion
T-pinch
Stack loss Fuel
Enthalpy Q-Hmin
Page 67 of 208
Another case as shown in the graph below when the shape of the grand composite curve (GCC) prevent the stack temperature to be reduced all the way to the pinch temperature where the process away from the pinch limits the slope of the flue gas line and hence the stack loss.
T* T*TFT
Flue Gas
T*stack
T-pinch T*ambient Stack loss Fuel Enthalpy Q-Hmin
Page 68 of 208
Combined thermal energy and power using GCC A more complex utility is combined heat and power or cogeneration. In such case, simply, the heat rejected by a heat engine such as steam turbine, gas turbine or diesel engine is used as the hot utility. According to the first law of thermodynamics, heat and power are related. Thus it should not be surprising that the energy integration concepts can be applied for heat and power integration. Fundamentally there are three possible ways to integrate a heat engine exhaust, but before starting to discuss these possibilities let us get a glance about heat engines. In this course we mean with heat engines steam turbines and gas turbines. A heat engine takes in heat (Q1) from a heat source at very higher temperature and rejects heat (Q2) at a lower temperature into a heat sink while producing electricity. The power generation, in form of work (W), equals the difference between heat input (Q1) and heat rejection (Q2), as per the strict definition of the first law of thermodynamics. Thermodynamic texts imply that the efficiency of a heat engine is always less than 100% since some of the heat output will always be wasted. If we can utilize this waste heat in another process task to satisfy certain process requirement such as the heating of process streams then we are enhancing the efficiency of the heat engine via its integration with other process unit. A systems approach now will lead to different conclusion from the consideration of a particular unit in isolation of the rest of the process. The challenge now is how we integrate the heat engine with other process units systematically, without enumeration and in a way that enhance the thermal efficiency of the heat engine. As we mentioned earlier there are three possible way of such integration with process units, above the pinch, below the pinch or across the pinch. Let us first consider the integration of the heat engine exhaust across the pinch as shown in the graph (A) below.
Page 69 of 208
(A)
T* QHmin
+ QHE
Heat Engine
QHE - W
Qc+QHE-W
Page 70 of 208
The heat engine takes in QHmin units of heat from the external utility above the pinch, produces W units of work and rejects QHE-W heat into the process below the pinch, which is rejected into cooling water. The graph shows that the process still needs QHmin and the cooling utilities need will be increased by (QHE- W) amount of heat and the heat engine will perform no better than if it was a standalone one. It takes QHE and produce W work while it produces QHE-W waste heat. As a matter of fact this integration will be counterproductive from both capital investment and energy point of views. More capital investment since the cascaded heat (QHE-W) is rejected to the cooling water therefore the process will need more heat exchangers. More energy since the level of work extraction from the QHE supplied heat was not full than it should be by taken the expansion in the turbine/Heat engine all the way down to the condensation level to generate more power. Therefore, we conclude that heat engines should not be integrated across the pinch since this will be counterproductive from energy consumption point of view and the heat engine efficiency as well. Now let us examine the other two possible cases of integrating the heat engine above the pinch as shown in graph (B) or below the pinch as per graph (C). In both cases the integration of the heat engine with the process will be beneficial in both cases as we will show below. First case is to study the effect of integrating the heat engine with the process above the pinch point.
Page 71 of 208
(B)
T*
QHmin + QHE QHmin- (QHE-W) QHE

Heat Engine
QHE-W
Qc
Page 72 of 208
If we integrate the heat engine above the pinch then we are rejecting the QHE-W exhaust heat to the process which is a heat sink above the pinch and already needs such external utility. This mechanism means less external utility by the amount of QHE-W or all the thermal energy input to the heat engine QHE is utilized; part as work and the wasted heat QHEW as a heating medium for the process. In other words we take in extra units of energy QHE than the QHmin required by the process and which we are going to take any way for a standalone heat engine and utilize it all via turning it into shaft work and process heating medium achieving 100% efficiency for the heat engine as if we are having a100% conversion of heat to power using the first law efficiency definition. Now if we integrate the heat engine below the pinch as shown in graph (C), we are not integrating the heat engine exhaust but we are taking waste heat from the process, which is a heat source below the pinch, and give to the heat engine to produce work. In such case we are getting double benefits, first, we are not getting external QHE for the heat engine or we are not increasing the hot utility requirements. Second we are also decreasing the cooling utility requirement by Qc-QHE. In other words, again we are getting to a situation that we are using the process waste heat below the pinch to generate electricity and saving cooling utility.
Page 73 of 208
(C)
T* QHmin
QHE
Heat Engine
Qc-QHE
QHE-W
Page 74 of 208
This situation of a system approach, compared to the unit based approach, shows that when we put our boundary lines around both the process plant and the utility plant and consider a bigger picture than that we all of us sometimes see we get into a new system in which the process and the heat engine are acting as a one system. In such new situation, apparently the conversion of heat to work is happening with 100% efficiency. Now let us take a closer look at the two most commonly used heat engines, the steam turbine and the gas turbine to see the efficiency that they can achieve in practice. Simple rule, before we start talking about steam and gas turbines, is that the grand composite curve for the process should be used to make any quantitative assessment of any heat and power scheme and the heat engine exhaust shall be treated as any other utility. Steam Turbine Integration
The graph below shows a steam turbine expansion on an enthalpy-entropy plot. It shows the ideal expansion process and the real expansion one and what we mean by isentropic efficiency or second law (of thermodynamic) efficiency.
Page 75 of 208
Steam Turbine Expansion
Enthalpy (H)
H1
P1
Real expansion Process P2
Ideal expansion Process Isentropic H2
X=1.0 H2* X=0.9
X=0.85
Entropy (S)
In an ideal turbine, steam with an initial pressure P1 and enthalpy H1 expands isentropically to pressure P2 and enthalpy H2. In such hypothesis, the ideal work produced is equal to H1-H2. Because of the losses in the turbine real expansion process due to frictional effects in the turbine nozzle and blade passages, etc the exit enthalpy is greater than it would be in the ideal process. Therefore, the actual work output will be less and determined by the H2* value as shown in the above graph. The actual work output is equal to H1-H2*. The turbine isentropic efficiency
isen = ( H 1 H 2*) /( H 1 H 2)
Page 76 of 208
The output from the turbine might be superheated or partially condensed. If the exhaust steam is to be used for process heating, ideally it should be close to saturation condensation. If the exhaust steam is significantly superheated, it can be desuperheated by direct water injection of the boiler feed water which vaporizes and cools the steam. However, if saturated steam is fed to a steam main, with significant potential of heat loss from the main, then it is desirable to retain superheat than to de-superheat. Since the heat losses will cause excessive condensation in the main, which is not desirable. On the other hand if the exhaust steam from the turbine is partially condensed, the condensate is separated and the steam is used for heating. Figure below shows a steam turbine integrated with process plant above the pinch.
Page 77 of 208
Fuel supplied to the boiler is used to produce high pressure steam for process heating and for power generation in the turbine. Rejected heat from the turbine is also used for process heating. As shown in graph the process proceeds as follows heat QHP is taken into the process from high pressure steam. The balance of the hot utility demand QLP is taken from the steam turbine exhaust. The flash steam is recovered after pressure reduction of the high-pressure steam condensate as shown in the above graph. Heat Qfuel is given to the boiler and an overall energy balance gives: Q-fuel = Q-HP steam+ Q-loss And Q-HP steam = QHP+ QLP+W; Then Q-fuel= (QHP+ QLP+ W) + Q-loss The process needs QHP+ QLP to satisfy its enthalpy imbalance above the pinch. In addition to boiler losses there are losses from the turbine and significant one from the steam distribution system.
Gas Turbine Integration
Graph below shows a simple gas turbine matched against a process. The gas heat engine is essentially a rotary compressor mounted on the same shaft of a turbine. Air enters the compressor where it gets compressed before entering the combustion chamber where there the combustion process increases its temperature. The mixture of the combustion gases, including air, is expanded in the turbine. The energy input to the combustion chamber is enough to drive the compressor and produce useful work (Electricity). The expanded gas may be discharged to the atmosphere directly or may first be used to preheat the air to the combustion chamber as shown in graph below.
Page 78 of 208
Gas turbines are normally only used for relatively large-scale applications. The overall efficiency of conversion of heat into power depends on the turbine exhaust profile, the pinch temperature and the shape of the process grand composite curve. It is instructive to mention here that reciprocating engines, combined cycles, etc., all have heat sources and heat sinks and they can be treated similarly. Since we have many different types of heat engines to choose from, we need the grand composite curve to help us determine the most suitable heat engine for a given process. Example in graph below shows that for a process that have such type of a grand composite curve the best type of utility will be low pressure steam. This result is not easy to obtain via intuition without drawing the process grand composite curve due to the fact that the process extends to high temperatures.
Page 79 of 208
The grand composite curve in the above case indicates a little bit flat temperature profile for demand which exhibit a good fit between low pressure steam and the process. If high pressure steam is available we can utilize the energy in it via producing work using steam turbine to recover the shaft work available between the high pressure (HP) and low pressure (LP) mains as shown in graph below.
Page 80 of 208
Integration of a steam turbine with the process

T
HP
W LP
120 C
In other situations where the process grand composite curve slope is far of being flat, integration of steam turbines with the process might not be appropriate and a gas turbine might be better as shown in graph below. The graph shows that a gas turbine is more appropriate because the sloping profile of the GCC is not suitable for a constanttemperature utility like steam. The gas turbine exhaust here in this example has a sloping utility profile which better fits the process GCC since several steam levels would be required to make a steam turbine efficient.
Page 81 of 208
In most of the times there is a trade-off between capital and energy costs. In the selection of steam or gas turbines that best integrate with the process utility requirements we have a trade-off between gas turbine efficiency, HEN capital cost and stack heat loss to ambient. For instance in order to closely matches the process grand composite curve with the gas turbine exhaust, the minimum approach temperature between the process and the hot utility exhaust gas will be small resulting in high capital cost for the HEN. In addition it will result in increased heat loss to the ambient. On the other hand higher stack temperature means less efficient gas turbine and lower flowrate-specific heat availability to satisfy the process needs. These two cases are depicted in the graph below. The solid line of the gas turbine exhaust represents a low temperature gas turbine exhaust, while the dotted line represents a higher gas turbine exhaust temperature.
Page 82 of 208
Integration of Heat Pumps Heat pumps are the opposite of heat engines. We put work into a heat pump to raise the temperature level of the available heat. A heat pump is a device that absorbs heat at low temperature in the evaporator, consumes shaft-work when the working fluid is compressed and rejects heat at higher temperature in the condenser. The condensed working fluid is expanded and partially vaporized. Then the cycle repeats. In many cases the working fluid is pure component which means that the evaporation and condensation take place isothermally.
Page 83 of 208
Graph below is a schematic of a simple vapor compression heat pump.
Schematic of a Simple Vapor Compression Heat Pump
Condenser
Expansion Valve
W
Compressor
Separator
Evaporator
Just as with heat engine integration there are appropriate and inappropriate ways to integrate heat pumps. Essentially there are two ways to integrate heat pumps with the process; either across the pinch or not across the pinch. Not across the pinch means the heat pump will be placed above the pinch or below the pinch point. Let us firs examine the case where the heat pump is placed or integrated with the process above the pinch point. In this configuration the process imports W shaft-work and saves W hot utility. In such configuration the system converts power into heat which is normally never economically worthwhile considering.
Page 84 of 208
(A)
Integration of Heat Pump above the Pinch
T*
QHmin _ W
(QHP+W) QHP
Heat Pump
Pinch
Qc-min
Page 85 of 208
Another possible integration not across the pinch too is integrated the heat pump with the process below the pinch as shown in the (B) graph below.
(B)
Integration of Heat Pump below the Pinch
T* QHmin
Pinch
QHP+ W
Heat Pump
QHP (Qc-min) + W
Page 86 of 208
The result of integration in such case is worse economically. In such case power is turned into waste heat that then need to be rejected to cooling utility. Last possible way of integration of the heat pump with the process is across the pinch as shown in graph below.
(C)
Integration of Heat Pump across the Pinch
T*
(QH-min)- QHP-W
QHP+ W
Pinch
Heat Pump
QHP
(QC-min)-QHP
Page 87 of 208
This arrangement brings about a genuine energy savings. As shown in graph it can reduces the process heating utility import by (W+QHP) and decreases the process cold utility import by QHP. It also makes overall sense because heat is pumped from the part of the process which is overall a heat source to the part of the process which is overall a heat sink. Thus the appropriate placement of heat pumps in the process context is that they should be placed across the pinch. In case of utility pinch the heat pumps shall also be placed across it. Graph below shows a heat pump integrated properly across the pinch and the use of grand composite curve in defining the best place to integrate this heat pump. The (A) graph depicts the heat balance but the (B) graph of the GCC illustrates how the grand composite curve can be used to size the heat pump.
The Heat Pump Against the GCC
(A) (QH-min)- QHP-W
(B)
T*
T* Steam
QHP+ W
QHP+W
Pinch
Heat Pump
QHP QHP CW (QC-min)-QHP Enthalpy
It is instructive to mention here that in order to examine how a heat pump performs we need to check its coefficient of performance (COP). The coefficient of performance for a heat pump (COP) generally can be defined as the useful energy delivered to the process divided by the shaft-work spent to produce this useful energy.
Page 88 of 208
COPHp = (QHp + W)/ W Where COPHp is the heat pump coefficient of performance, QHp is the heat absorbed at low temperature and W is the shaft-work consumed. For any given type of heat pump, a higher COPHp leads to better economics. Having a better COPHp and hence better economics means working across a small temperature lift with the heat pump. The smaller the temperature lift, the better is the COPHp. For most allocations, a temperature lift greater than 25C is rarely economical. Attractive heat pump application normally requires a lift much less than 25C. Using the grand composite curve, as shown above, the loads and the temperatures of the cooling and heating duties and hence the COPHp of integrated heat pumps can be readily assessed. To target for the thermal design of the heat pump we can use the GCC and treat the heat pump like an energy utility. The graph below shows that the GCC can help define the W needed for the heat pump if the process cooling utility Qc and the desired temperature differences, which can bring up value, are defined. Since the W needed in the heat pump is a function in the temperature difference; iteration will be needed to decide best temperature difference. In the graph below it is clear that the hot side of the heat pump is limiting since the desired process heating utility Qh from the heat pump is limiting.
The Heat Pump Against the GCC
T* Steam
With the GCC determining the levels for heat pump placement is straightforward
QHP+W W
Heat Pump
QHP
CW Enthalpy
Page 89 of 208
Let us try to go back to the coefficient of performance (COP) of a heat pump and try to see in numbers how it is function of temperature difference. We know that the shaft-work requirement of a heat pump increases with T .If the heat pump operates over small T the shaft-work requirement is lower than working at high T and consequently as shown in the graph below the COP is relatively becomes high.
Use Heat Pump Across Small Ts COP = (QHP+W)/W T* T*
QHP+W=150 W=50
Heat Pump
QHP+W=110 W=10 QHP=100

Heat Pump
QHP=100
COP = 150/50=3
Enthalpy COP = 110/10=11 Enthalpy
Integration of Refrigeration Systems A refrigeration system is a heat pump in which heat is absorbed below ambient temperature. A heat pump is a reverse of a power cycle. For example a home refrigerator removes heat from food that is just above freezing (say 5C) and ejects that heat into the room which is at ambient temperature( say 25C). The work we put into the pump to move the heat to the higher temperature degrades to heat. Degrading heat from a high temperature to a low temperature allows us to create work. Using work allows us to elevate the temperature of low grade heat. Before we talk about the integration of refrigeration cycles with the process let us examine first the refrigeration cycle using temperature-entropy diagram.
Page 90 of 208
The graph below shows again the main components of a typical refrigeration cycle. We start examining the cycle at the exit of the condenser at point # 1.
A Typical Refrigeration Cycle
High Pressure Liquid
1 4
Condenser
High Pressure Vapor
Expansion Valve
W
Compressor
2
Low Pressure Vapor and Liquid (two Phases) Evaporator
3
Low Pressure Vapor
Here the refrigerant is a high pressure liquid, very near to saturation (i.e. about ready to boil). We reduces the pressure on the liquid by passing it through an adiabatic valve (H=0.0). It partially vaporizes point #2. The heat required for vaporization, since we do not give it an external heat, comes from the fluid itself, cooling it. We next pass this fluid through the refrigeration coils where the rest of the liquid evaporates. In doing so, it takes heat from the surroundings (from food or process). We now have a low pressure liquid/fluid, point # 3, which is all vapor and very near saturation (just ready to condense). We then increase the pressure on the fluid by compressing it. An ideal compressor operates isentrpoically (at constant entropy, S=0.0), arriving to point # 4. It has been heated up due to compression becoming a superheated vapor well above saturation. We then cool it by rejecting the heat to the surrounding or cooling medium in the process returning ultimately to being a liquid at high pressure, point # 1.
Page 91 of 208
The graph below shows this cycle on a plot of temperature versus entropy diagram. The advantage of viewing such cycle on temperature versus entropy and not pressure versus enthalpy that the area enclosed in the cycle represents the ideal work needed to run the cycle. Any improvements to the cycle will show up as reductions in this area, provided that we pick up the same amount of heat in the evaporator both before and after the improvement since this amount of heat is normally the one dictated by the process needs.
Temperature- Entropy Diagram for refrigeration Cycle
T, K 4
liquid
1 Vapor 2
Vapor and liquid
Entropy, S(J/mol K)
Understanding the above temperature- entropy diagram in the context of the refrigeration cycle components allows us to generate ideas that can help save energy. In the graphs below we will present two of such possible improvements.
Page 92 of 208
The first is to use a multi-stage compressor as shown below to reduce the area enclosed in the cycle which represents the idea work needed to run the cycle.
T, K
Save this area
liquid Vapor Multistage Compressor
Vapor and liquid
Entropy, S(J/mol K)
In such case we compress only part way and then cool the vapor back to its saturation temperature. We compress again to the final pressure. The area saved on the right side of the above graph represents the savings in the ideal work needed to run this cycle.
Page 93 of 208
The second possible improvement shown in the graph below is using a let down turbine rather than a valve to drop the pressure of the high pressure liquid.
T, K
Turbine Expansion Save this area
liquid Vapor Multistage Compressor Vapor and liquid

Extra refrigeration capacity
Entropy, S(J/mol K)
This expansion is shown in the straight line. This step appears to increase the area enclosed which means extra ideal work will be required to run this cycle. However, it also increases the length of the line that represents the heat we pick up in the evaporator from the process. In other words we are increasing our refrigeration capacity. It is really an improvement since the area (that represent the ideal work needed to run the cycle) per unit heat we pick from the process (process demand) in the evaporator is actually reduced when we use the let down turbine. In general we should normally use one cycle to elevate the low temperature heat by no more than30 C. If we need to increase the temperature of the heat more than that, it pays to use multiple cycles where a lower temperature cycle passes heat to the cycle above it, which in turns passes the heat to the cycle above it, repeating until the top cycle, which passes the heat to the ambient conditions. This configuration is shown in the double cycle shown in the graph below.
Page 94 of 208
A Two Stage Refrigeration Cycle
Condenser
Expansion Valve
Evaporator
Compressor
Expansion Valve
Condenser Compressor
Evaporator
Refrigeration cycles are expensive to purchase and expensive to operate. Hence they should be run with much smaller driving forces than are typical for above ambient processes. Smaller driving forces mean we will pay more for the equipment but less for operating costs. The evaporator/condenser that connects the two cycles in graph above requires a temperature driving force for the heat to transfer. The lower cycle must raise the heat to a temperature just above the temperature of the fluid in the upper cycle so it can transfer heat to it. If it is reasonable to use the same refrigerant in the two cycles we can eliminate this loss in temperature driving force by exchanging heat between the two cycles as shown in the graph below without this common evaporator/condenser shown in the graph above.
Page 95 of 208
A Two Stage Refrigeration Cycle
Condenser
Expansion Valve Two phase fluid
Flash
Vapor
Compressor
Liquid
Compressor Expansion Valve
Evaporator
In this graph we replace the evaporator/condenser unit with flash drum. The two cycles trade fluid rather than just heat. The lower cycle puts vapor into the flash unit while the upper cycle feeds in 2-phase fluid. The lower cycle takes away the liquid while the upper cycle takes the vapor from the flash unit. Material balance requires each cycle to remove the same amount of refrigerant as it put into the flash unit. The lower cycle trades vapor for liquid, while the upper cycle trades vapor and liquid for vapor alone. It is as if they have traded heat. This trade is done with no temperature driving force and makes it an attractive alternative to improve a cascaded refrigeration cycle. There are many other ways to improve refrigeration cycles and what does really concerns us is to design a good refrigeration cycle that also best fit with our process using the type of insights that we got from grand composite curve.
Page 96 of 208
Integration of Refrigeration Cycles As we mentioned before the refrigeration system is a heat pump in which heat is absorbed below ambient temperature. Thus the appropriate placement principle for heat pumps applies in exactly the same way as for refrigeration cycles. The appropriate placement for refrigeration cycles is that they also should be across the pinch. As with heat pumps refrigeration cycles can also be appropriately placed across utility pinches. It is common for refrigeration cycles to be placed across a utility pinch caused by maximizing cooling water duty. Most refrigeration cycles are almost the same as heat pumps. Heat is absorbed at low temperature, servicing the process, and rejected at high temperature either to the ambient (cooling water or air coolers) or to heat recovery in the process. Heat transfer takes place essentially over latent heat profiles. As with heat pumping the GCC is used to assess how much heat from the process needs to be extracted into the refrigeration system and where if appropriate the process can accept rejected heat.
The Heat Pump Against the GCC for Appropriately Placed Heat Pump
T*
(QH-min)- QHP-W
T* Steam
QHP+ W
QHP+W
Pinch
Heat Pump
QHP QHP CW (QC-min)-QHP Enthalpy
Page 97 of 208
Again the coefficient of performance COP determines how the refrigeration system performs. In the case of refrigeration systems, the coefficient of performance COPref is generally defined in terms of the heat extracted from the process divided by the shaftwork consumed. COPref = QHP/ W The higher the coefficient of performance the better is the economics. The cost of the shaft-work required to run a refrigeration system can be estimated approximately as a multiple of shaft-work required for an ideal system. The performance of an ideal system is given by Wideal/Qc = (Th-Tc)/ Tc Where Wideal= Ideal shaftwork required for the refrigeration cycle Qc = The cooling Duty Tc= Temperature at which heat is taken into the refrigeration cycle(K) Th= Temperature at which heat is rejected from the refrigeration cycle(K) The ratio of ideal work to actual work is usually 0.6. Thus the actual work W required for the refrigeration cycle will be; W= (Qc/0.6)* ( Th-Tc)/Tc
Page 98 of 208
Before we leave heat pumps and refrigeration cycles we need to refer to a technique used in enhancing energy consumption in stand-alone distillation columns know as vapor recompression. The graph below shows what we mean by vapor recompression.
Distillation Column with Vapor Compression Scheme
Vapor compression or some people sometimes call it recompression is an expensive way to save energy in distillation columns. It uses a compressor which has a high capital cost and operating too. It shall be avoided as a technique for energy saving in distillation process unless the column is already placed across the pinch with high reboiler and condenser duties besides that the difference in temperature between the Treboiler and T-condenser is not much. It is the same philosophy which governs the case where we decide the economics of a heat pump. The heat temperature upgrade should not be high otherwise the economics will not allow it.
Page 99 of 208
Heat Integration of Distillation Columns The dominate heating and cooling duties associated with a distillation column are reboiler and condenser duties. In general there will be other duties associated with the heating and cooling of feed and product streams. These sensible heat streams will be smaller in comparison with the latent heat changes in reboilers and condensers. Both the reboiling and condensing processes normally take place over a range of temperature. Practical considerations, however, usually dictate that the heat to the reboiler must be supplied at a temperature above the dew point of the vapor leaving the reboiler and that the heat removed in the condenser must be removed at a temperature below than that of the bubble point of the liquid. Therefore, in preliminary design at least both reboiling and condensing can, be assumed to take place at a constant temperature. As we did in heat pumps we will examine here the consequence of placing the distillation columns in different locations relative to the pinch. There are two possible ways in which the distillation column can be integrated with the rest of the process. The reboiler and the condenser can be integrated either across the pinch or not across the pinch. In such case where distillation is going to be placed across the pinch and the background process (which does not include the reboiler and condenser) is represented simply as a heat sink and heat source divided by the pinch. Heat Qreb is taken into or given to the reboiler above pinch temperature and rejected from the condenser at a lower temperature, which in the graph shown below is below pinch temperature.
Page 100 of 208
Integration of Distillation Column across the Pinch
T*
(QH-min)+Qreb
Distillation Column is upside down
Qreb
Pinch
Distillation Column
Qcond
(QC-min)+Qcond
Because the process sink above the pinch requires at least Qh-min to satisfy its enthalpy balance, the removal of Qreb into the reboiler must be happening by introducing an extra Qreb from hot utility. Please note here that if the distillation column will be in stand-alone mode the heating utility requirement for the process and the distillation column will be the same in case of integration between the process and the distillation column and without integration. Below the pinch the process is a source and needs to reject heat to the environment (cooling utility) to reach the enthalpy balance below the pinch and the rejection of heat through the condenser of the distillation column will add up to the cooling duty requirement. Therefore, it is clear from the graph below in such case of integration between the distillation column and the process that fundamentally there is no energy saving from the integration of a distillation column across the pinch.
Page 101 of 208
In integration of the distillation column and the process above or below the pinch the situation is different. The graph below shows a distillation column entirely placed above the pinch.
Integration of Distillation Column above the Pinch
T*
QHmin+ Qreb-Qcond
Qreb
Distillation Column
Note: Distillation Column is Shown in an upside down position
Qcond
Pinch
Qc-min
Page 102 of 208
The distillation column takes heat Qreb from the process and returns Qcond at temperature above the pinch. The hot utility consumption changes by (Qreb-Qcond). The cold utility as shown in graph is unchanged. Usually Qreb is equal to Qcond, then the heat utility consumption is QH-min, and there is no additional hot utility required to run the column. In such case the integration between the process and the column is useful. In graph below, the last possible way of integration between the column and the process where the column is integrated with the process below the pinch is shown. Now the hot utility is unchanged, but the cold utility consumption will change by Qcond- Qreb and hence they are almost equal as we said before, then the result is similar to heat integration above the pinch.
Integration of Distillation Column below the Pinch
T* QHmin
Distillation Column is upside down
Pinch
Qreb
Distillation Column
Qcond (Qc-min) + Qcond- Qreb
Page 103 of 208
All the above arrangements can now be described in a very simple statement: The appropriate placement for distillation column should not be across the pinch. It is instructive to note here that due to controllability/operability reasons integrating both the reboiler and the condenser with the process can make the column difficult to control and start-up. The appropriate placement principle can only be applied if the process has the capacity to give up or accept the required heat duties in the distillation column. A quantitative tool is therefore needed to assess the source and sink capacities of any given background process. For this purpose the grand composite curve is used. The grand composite curve would contain all heating and cooling duties for the process including those associated with the column feed heating and product cooling but excluding reboiler and condenser loads. Let us consider now few examples for the use of the grand composite curve in helping find the appropriate place for the heart integration of the distillation column with the process. In the two graphs below the grand composite curve is shown and both the distillation column reboiler and condenser duties are shown too separately and matched against the GCC. Neither of the distillation columns in the two graphs fit with the GCC. In the first graph it is quite clear the distillation column exists across the pinch. In the second graph even though the distillation column reboiler and condenser duties lay above the pinch it also does not fit with the GCC. Strictly speaking, it is not appropriately placed and hence some energy can be saved.
Page 104 of 208
Distillation Column Which Do Not Fit Against the Grand Composite Curve T*
Reboiler
Condenser
Enthalpy
Page 105 of 208
Distillation Column Which Do Not Fit Against the Grand Composite Curve T*
Reboiler
Condenser
Enthalpy
Page 106 of 208
By contrast the distillation column shown in graph below fits. The reboiler duty can be supplied by the hot utility. The condenser duty must be integrated with the rest of the process
Distillation Column Which Fit Against the Grand Composite Curve T*
Reboiler
Condenser
Enthalpy
Another example is shown in graph below. This distillation column also fits. The reboiler duty must be supplied by the integration with the rest of the process. Part of the condenser duty must be integrated but the remainder of the condenser duty can be rejected to the cold utility. The design of simple distillation columns that have good impact on utility consumption can be evolved from heat integrated processes using the GCC.
Page 107 of 208
Starting with an inappropriately placed distillation, if it is shifted above the pinch by increasing its pressure, the condensing stream which is hot stream, is shifted from below to above the pinch. The reboiling stream, which is a cold stream, stays above the pinch. If the inappropriately placed distillation, is shifted below the pinch by decreasing its pressure, then the reboiling stream, which is a cold stream is shifted from above to below the pinch. The condensing stream which is hot stream stays below the pinch. Thus the appropriate placing of distillation column is in fact a particular case of the plus/minus principle.
Distillation Column Which Fit Against the Grand Composite Curve T*
Enthalpy
Page 108 of 208
If a distillation column is in appropriately placed across the pinch, it may be possible to change its pressure to achieve appropriate placement. Of-course, as the pressure is changed the shape of the distillation column reboiler and condenser box will also be changed.(since not only do the reboiler and condenser temperatures change but also the difference between them). The relative volatility also will be affected. Thus both the height and the width of the box will change as the pressure changes. Changes in the columns pressure will also affect the heating and the cooling duties for column feed and products. Since these streams are also included in the background of the process the GCC will also change and the fit between the GCC and the distillation column is going to be different than the one before the modification. If the distillation column will not fit either above or below the pinch, then other design options can be considered. One possibility is the double effecting scheme shown below.
Double-Effect Distillation Column
C1
The relative pressures in the two columns are chosen in a way that allow best integration with the process
C2
Page 109 of 208
The column feed is split and fed to two separate parallel columns. The relative pressures in the column are chosen such that the two columns can each be appropriately placed. Obviously, the capital cost of such a scheme will be higher than that of a single column, but it may be justified in some favorable conditions.
Double-Effect Distillation Column
T*
C2
C1
Enthalpy
Page 110 of 208
Another design configuration that can be considered if a column will not fit with the process GCC is use of an intermediate reboiler or condenser. An intermediate condenser is shown in the graph below.
Distillation Column with Intermediate Condenser

A
Condenser
cooler
A+B
Reboiler
Page 111 of 208
Distillation Column with Intermediate Condenser
T*
Reboiler
Condenser
Intermediate Condenser
Enthalpy
The shape of the box is now altered because the intermediate condenser changes the heat profile of the column. The particular design shown above in the GCC diagram would require that at least part of the heat rejected from the intermediate condenser be passed to the process. An analogous approach can be used to evaluate the possibilities for use of intermediate reboilers. It is important to note here that such heuristics are not always correct and needs to be used with care.
Page 112 of 208
Heat pumping (vapor recompression) schemes sometimes are also considered in stand-alone distillation column designs for efficient use of energy. For heat pumping to be economical on a stand-alone basis, the heat pump must work across a small temperature difference, which for distillation means close-boiling mixtures. Sometimes the scope of integrating simple distillation columns is limited. Some constraints prevent integration of distillation columns with the rest of the process. Considering the sequence of the distillation in the context of the whole process from heat integration point of view is one of the options that can be pursued in such cases. Some other option in looking for energy efficient process with distillation columns is the consideration of more complex distillation column designs. The design changes suggested to distillation columns will always be motivated by the requirement to reduce energy costs by more effective heat integration between the distillation column and the rest of the process. There are however capital cost implications when distillation column are integrated with the rest of the process. These implications fall into two categories, changes in column capital cost and changes in heat exchanger network capital cost. Obviously, these capital cost changes should be considered together with the energy cost changes in order to achieve an optimal tradeoff between capital and energy costs. Distillation column capital costs: The classic example of optimization in distillation columns is the tradeoff between the capital cost of the column and the energy cost for the distillation. This would be carried out with the distillation column working on utilities for both the reboiler and condenser and not for a column integrated with the rest of the process. The column optimal reflux ratio is normally determined according to the tradeoff between the column capital cost and the column operating cost. An increase in the reflux ratio can result in lower number of trays and hence smaller in length column. On the other hand such column will have more loads on the condenser and reboiler resulting in bigger size and more utilities cost. If the column is appropriately integrated with the rest of the process, then the reflux ratio often can be increased without changing the overall energy/operating cost of the column. Having said that it is important to note that the decrease in the capital cost of the column is not the only capital cost that should be explored to find the optimal reflux ratio. In order to find the true tradeoff, the HEN capital cost should be evaluated too.
Page 113 of 208
Heat exchanger network capital cost: The full implications of the integration between the distillation columns and the rest of the process are only clear by considering the composite curves of the process with the distillation column. Temperature driving forces become smaller through out the process due to the process conditions modifications that enable the best fit of the distillation columns with the rest of the process. This means that the capital energy tradeoff should be re-adjusted and a larger Tmin may be required. The optimization of the capital/energy tradeoff might undo part of the savings achieved by appropriate integration. 2.4 Heat Exchangers Network (HEN) Design
Fewest-Number of Units A heuristic estimate for the minimum number of units is obtained by using Euler's Rule from Graph Theory as the basis: U=N+LS where U is the number of units (process/process heat exchangers, heaters and coolers), N is the total number of process streams and utility types, L is the number of heat load loops in the network and S is the number of sub-systems in the network. Assuming there are no heat load loops (it will be shown later that loops can be removed) and no sub-networks (sets of hot and cold streams in perfect heat balance, which would be a coincidence), the following can be used as an estimate for the fewest number of units: Umin = N + 0 - 1 = N 1 In order to obtain Maximum Energy Recovery (MER) or minimum energy consumption, however, it was shown above that decomposition at the Process Pinch must be respected. This means that separate heat exchanger networks must be designed above and below Pinch, and the corresponding minimum total number of units is given by: Umin, MER = (N - 1 ) above + ( N - 1 )below
Page 114 of 208
In the case of multiple utilities, as indicated in figure below, new Utility Pinch points will be introduced, whenever a cheaper utility is maximized in order to minimize a more expensive utility. There are also cases with near-Pinches that could be included to make tight design situations easier. A more general formula for the fewest number of units is function of number of Pinch points (Process and Utility Pinches) and the total number of process streams and utility types that are present between two neighboring Pinch points, alternatively above the highest Pinch and below the lowest Pinch. Stream Grid for the process example with Multiple Pinch Points
Utility Pinch
200 C 250 C
Process Pinch
180 C
Utility Pinch
170 C
HP
270 C
MP
H1
220 C
160 C
H2
210 C
60 C 50 C
C1
160 C 150 C 15 C
210 C
C2
180 C 160 C
LP
150 C
20 C
CW
Page 115 of 208
It is obvious that the target for minimum number of units depends on the number of utility types that are used and the number of Pinch points (process and utility pinches) where strict decomposition is implemented. For our simple example, with 4 process streams, (keeping the distillation column out of the discussion) and up to 4 utility types, the fewest number of units varies considerably. If we only use HP and CW and do not decompose at the process Pinch, the fewest number of units is 4, while it is 14 if we use all 4 utility types and decompose at all 3 Pinch points. Obviously, the economic trade-off between energy cost and equipment cost will have an optimum that is closer to 5 heat transfer units than 14. The Stream Grid (Linnhoff and Flower, 1978a) shown above is an important representation for the design of heat exchanger networks. It can also be used to assist in the application of the (N-1) rule to calculate the fewest number of heat exchangers for the various scenarios of multiple utilities and the existence of Process and Utility Pinch points. Minimum Number of Shells Refinements have been made in Pinch Analysis (Ahmad and Smith, 1989) to reflect the fact that very few industrial heat exchangers are pure counter-current. These refinements relate to both the number of heat exchange units (now counted as number of shells rather than heat exchangers) and to heat transfer area. So far, these extensions only apply to Shell & Tube exchangers, where correction factors for heat transfer area (fT) are used that depend on mCp values and temperatures for the streams. These factors represent deviations from pure countercurrent heat exchange when using models and equations for 1-2 Shell & Tube exchangers. If the value of fT falls under a minimum acceptable value, the number of shells must be increased by one, and the procedure is repeated. With these extended models, it is possible to obtain a target for the minimum number of shells rather than units. The next section on minimum heat transfer area also applies to shells in 1-2 configurations, with the addition of the fT factor when calculating area. Minimum Heat Transfer Area Estimating the need for total heat transfer area in the network of heat exchangers before design is both the most time consuming (need software) and the most uncertain targeting activity. There are large uncertainties in heat transfer coefficients, and simplified assumptions are made about the network structure when calculating minimum total area. In Pinch Analysis, a target for minimum area is obtained by applying and expanding the concept of counter current heat exchange between two streams to the situation with many hot and cold streams. The resulting heat flow model is the vertical one illustrated in figure below. The idea of Vertical Heat Transfer between the Composite Curves is aiming at optimal use of the available driving forces in order to minimize total heat transfer area.
Page 116 of 208
Since, however, the general equation for heat transfer area is: A = Q / ( U TLM fT ) it is the product of heat transfer coefficient (U) and driving forces (TLM) that should be optimally distributed, not driving forces alone. This will be briefly discussed below. Vertical Heat Transfer for the process Example:
T (C)
HP
CW
Q (kW)
In order to achieve vertical heat transfer in a heat exchanger network, however, all heat Exchangers in the same Enthalpy Interval must have exactly the same temperature profiles. This can only be achieved by considerable splitting and mixing of streams and a large number of small heat exchangers (must apply the N-1 rule to each enthalpy interval). The corresponding network is therefore referred to as the "Spaghetti Design", and serves exclusively as a calculation model for total heat transfer area (Townsend and Linnhoff, 1984).
Page 117 of 208
Hot and cold utilities are included and the result is often referred to as the Balanced Composite Curves. In this case only HP steam and cooling water are used. The actual calculation of minimum area, based on the concept of vertical heat transfer, is done with the so-called Bath formulae (after the place where the equation was presented): Amin =
(1 / Tlm, j )i (qi ) /( hi )
Where qi is the change in enthalpy and hi is the film heat transfer coefficient for stream (i) in enthalpy interval (j). By applying this equation to the example problem we get the following results: With MP/LP: Amin = 775 m Without MP/LP: Amin = 632 m Thus, while the introduction of MP and LP steam reduces total energy cost with about 30%, there is an increase in the target for heat transfer area of about 23%. In addition, as discussed in the previous section, there will be an increase in the number of units, and the network structure will be more complex. As mentioned above, there are a number of uncertainties related to these target values for minimum area. In addition to the fact that heat transfer coefficients are uncertain by nature; the vertical model and the Bath equation have two severe limitations: To achieve minimum area, a large number of heat exchangers; splitters and mixers are required. Due to economy of scale effects, cost optimal heat exchanger networks will have close to the fewest number of units rather than close to minimum area. The so called Spaghetti Design should only be regarded as a model for calculating Amin. The strict vertical model will only result in minimum area if all film heat transfer coefficients for the hot streams area equal (hH), and that all cold stream film heat transfer coefficients are equal (hC). With significant differences in these coefficients, streams with low film heat transfer coefficients should be matched with other streams while allowed more driving forces at the expense of matches between streams with large film heat transfer coefficients. The later streams shall be assigned less driving forces. As a result, there may be considerable non-vertical (Criss-Cross) heat transfer. These limitations are important, however, the main use of the target for minimum area is to be able to estimate total annual cost ahead of design for various values of Tmin, in order to identify a good starting point for the design exercise.
Page 118 of 208
Total Annual Cost By combining targets for minimum energy consumption, least number of units or shells and minimum heat transfer area, as well as cost data for utilities, cost equations for heat exchangers, some economic factors such as payback time or interest rate, and the number of operating hours per year, it is possible to obtain figures for Total Annual Cost. There are also uncertainties in these estimates, for example related to the fact that we only have figures for total area, and not how this area is distributed among the heat exchangers. With an economy of scale type cost equation, such a distribution is important for the final result. Experience from industrial projects have shown, however, that some of the uncertainties and assumptions in the calculation of area and total annual cost tend to cancel, and that the estimated total cost often is within a few percent from the total cost of the final heat exchanger network (using the same cost and economic data). As mentioned above, the main purpose of estimating Total Annual Cost (TAC) is to identify a good starting point for network design. This is done by calculating the different targets and the resulting total annual cost for various values of Tmin. By selecting a value for Tmin where TAC has a minimum, the initial heat exchanger network will have a structure that is compatible with the final optimal network. In case of multiple utilities a similar economic trade-off should be used to explore the cost impact during the targeting phase. Methods have been developed within Pinch Analysis that can be used to identify nearoptimal amounts of the various utilities (Parker, 1989, and Hall et al., 1992).It should also be mentioned that utility selection and process modifications interact and must be considered simultaneously. HEN Design This part will be presented in much less detail than the previous ones where number of concepts, representations and graphical diagrams were introduced that are of a general nature with several different applications in Process Integration. Design of Heat Exchanger Networks in various industries is primarily carried out using the now classical Pinch Design Method (Linnhoff and Hindmarsh, 1983). While the original method focused on minimum energy consumption and the fewest number of units, later graphical and numerical additions made it possible also to consider heat transfer area and total annual cost during design. Both the original features and the later extensions have been implemented in current state of the art commercial software packages for Heat Exchanger Network Design.
Page 119 of 208
The basic Pinch Design Method respects the decomposition at Process and Utility Pinch points and provides a strategy and matching rules that enable the engineer obtaining an initial network that achieves minimum energy consumption. The Stream Grid presented in before is very useful in the design phase and acts as a drawing board, where the engineer places one match at a time using these matching rules. The Pinch Design Method also indicates situations where stream splitting is required to reach the minimum energy target. Stream splitting is also important in area considerations and the optimal use of temperature driving forces. The design strategy mentioned above is simply to start design at the Pinch, where driving forces are limited and the critical matches for maximum heat recovery must be selected. The matching rules simply ensure sufficient driving forces, and they attempt to minimize the heat exchangers number of units. The design then gradually moves away from the pinch, making sure that hot streams are utilized above Pinch (limited resource), and vice versa for cold streams below Pinch (limited resource). The matching rules for Pinch exchangers (those situated immediately above or below Pinch) can be expressed mathematically by (where Hi and Cj are potential streams to be matched in a heat exchanger): Above Pinch: mCpcj mCphi nc nh Below Pinch: mCphi mCpcj nh nc Making sure that every unit fully satisfies the enthalpy change of either the hot or the cold stream (the tick-off rule) minimizes the number of units. If the inequalities above are not satisfied for a complete set of Pinch exchangers, stream splitting has to be considered in order to reach Maximum Energy Recovery (MER). It is always possible by stream splitting to satisfy all the inequalities, since total mCp for cold streams are larger than total mCp for hot streams above Pinch, and vice versa below Pinch.
Page 120 of 208
Find here below the detailed step-by-step pinch design method for the HEN synthesis. The best design for an energy efficient heat exchange network will often result in a tradeoff between the equipment and operating cost. This is dependent on the choice of the DTmin for the process. The lower the DTmin chosen, the lower the energy costs, but conversely the higher the heat exchanger capital costs, as lower temperature driving forces in the network will result in the need for greater area. A large DTmin on the other hand will mean increased energy costs as the will be less overall heat recovery, but the required capital cost will be less. The trade-off is further complicated in a retrofit situation, where a capital investment has already been made. Early in this course we leaned how to set energy and area targets for the process before considering the HEN design. In the early days of pinch technology this technique was important to help make the trade-off between the HEN capital cost and operating cost quickly and without any heavy calculations. However, nowadays lot of software are available to make a preliminary synthesis of any large size HEN and estimate its capital cost directly and then automatically make the trade-off between the operating cost and the capital cost for the HEN in order to determine the optimal Dtmin for the HEN to be designed. In this chapter before we get into the pinch design method procedures we will revisit the pinch targeting methods for since these methods can still be used using a pencil and paper enabling us to do some short-cut calculations quickly. First, the minimum number of units in a HEN (Nunits ) can be calculated using the following formula: Nunits = S 1 Where, S= number of streams (hot and cold) including utilities Then, the composite curves that make it possible to determine the energy targets for given value of DTmin can also be used to determine the minimum heat transfer area required to achieve the energy targets. To calculate the network area from the composite curves, utility streams must be included with the process streams in the composite curves to obtain the balanced composite curves. The resulting balanced composite curves should have no residual demand for utilities. Then the balanced composite curves are divided into vertical enthalpy intervals as shown in the graph below to calculate the total minimum area targets assuming constant overall heat transfer coefficient and pure vertical counter current heat transfer.
Page 121 of 208
Network Area Targeting

T (C)
HP
Qi
A3
A2
A1
A4
CW
A6
A5
Q (kW)
Ai = Qi / U TLMi A = i A i
Now to start the design of the HEN a good initialization of this design is to assume that no individual heat exchanger will have a temperature difference smaller than Tmin calculated from the targeting phase and there must be no heat transfer across the pinch by process to process heat transfer or/and inappropriate use of utilities. These rules are important for the HEN design to achieve the energy target, given that no individual exchanger should have a temperature difference smaller than Tmin. To comply with these two guidelines the design problem needs to be divided at the pinch and using the grid diagram as shown earlier in this course manual.
Page 122 of 208
Step-By-Step
HEN Design Method of a four streams problem The graph below shows the stream data of a HEN problem drawn on a grid diagram where the pinch temperature is shown on both the hot and the cold sides.
The Grid Diagram for the Step-By-Step HEN design

Pinch
150 C
CP (MW/C)
QHmin = 7.5 MW
250 C 200 C 180 C 230 C 140 C 40 C 80 C 20 C
2 4
0.15 0.25 0.2 0.3
3
140 C
QCmin = 10.0 MW
Start at the Pinch The pinch is the most constrained region of the problem. At the pinch, Tmin exists between all the hot and cold streams. As a result, the number of feasible matches in this region is severely restricted. Quit often there are essential matches to be made. If such matches are not made, the result will be either use of a temperature differences smaller than Tmin or excessive use of a utility resulting from heat transfer across the pinch. If the design was started away from the pinch, at the hot
Page 123 of 208
end or the cold end of the problem, then initial matches likely would need follow up matches as the pinch is approached, which violate the pinch or the Tmin criterion. Using the reverse logic and putting the argument the other way around, if the design is started at the pinch, initial decisions are made in the most constrained part of the problem. This is much less likely to lead to difficulties later. The CP inequality for individual matches The graph below shows the temperature profile for an individual exchanger at the pinch, above the pinch.
Criteria for Pinch Matches Above the Pinch

Pinch
150 C
CP (MW/C)
QHmin = 7.5 MW
250 C 200 C 180 C 230 C
2 4
150 C 150 C 140 C 140 C
0.15 0.25
1
0.20 0.30
3 140 C
Page 124 of 208
Moving away from the pinch, temperature difference must increase. Graph below shows the match between a hot stream and a cold stream which has a CP smaller than the hot stream. At the pinch the, the match starts with a temperature difference equal to Tmin. The relative slops of the temperature-enthalpy profiles of the two streams mean that the temperature differences become smaller while we are moving away from the pinch, which is infeasible.
T
Pinch
4 Hot stream
For hot stream #4 slope= 1/CP=1/0.25=4 For cold stream #1 slope=1/CP=1/0.2=5

Infeasible
T-min 1
Enthalpy
Page 125 of 208
On the other hand, the graph below shows the matching between the same hot stream (stream #4) but with another cold stream. (stream # 3)

Pinch
150 C
CP (MW/C)
QHmin = 7.5 MW
2 4
250 C 200 C 180 C 230 C
150 C 150 C 140 C 140 C
0.15 0.25
1
0.20 0.30
3 140 C
Page 126 of 208
The cold stream # 3 in this case has a greater CP as shown in graph. In such case the relative slopes of the temperature-enthalpy profiles now cause the temperature differences to become larger moving away from the pinch, which is feasible. Thus starting with Tmin at the pinch, for temperature difference to increase while moving away from the pinch, we have to have this inequality achieved CPH <= CPC (Above the pinch for streams at the pinch)
T
Pinch
4 Hot stream
For hot stream #4 slope= 1/CP=1/0.25=4 For cold stream #3 slope=1/CP=1/0.3=3.3

feasible
T-min 3
Enthalpy
Page 127 of 208
The graph below is depicting the situation below the pinch at the pinch. It shows the cold stream#1 matched with the hot stream #2 that has a smaller CP.
Criteria for Pinch Matches Below the Pinch

Pinch
150 C
CP (MW/C)
150 C 150 C 140 C
40 C 80 C 20 C
0.15 0.25 0.20
140 C
Page 128 of 208
In such cases if a cold stream is matched with a hot stream with a smaller CP, steeper slope, then the temperature differences become smaller which is infeasible as shown in graph below.
Pinch 2 1 T-min
For hot stream #2 slope= 1/CP=1/0.15=6.67 For cold stream #1 slope=1/CP=1/0.2=5
Infeasible Hot stream
Enthalpy
Page 129 of 208
If the same cold stream#1 is matched with a hot stream with a larger CP such as stream# 4 as shown in graph below different situation will arise.

Pinch 150 C
CP (MW/C)
150 C 150 C 140 C
40 C 80 C 20 C
0.15 0.25 0.20
140 C
Page 130 of 208
A less steep slope will be obtained resulting in temperature differences that become larger which is feasible as shown in the graph below. Thus starting with Tmin at the pinch, for temperature difference to increase while we are moving away from the pinch we have to have the following inequality achieved CPH >= CPC ( below the pinch for streams at the pinch)
T
Pinch
4 Hot stream 1 feasible T-min
For hot stream #4 slope= 1/CP=1/0.25=4 For cold stream #1 slope=1/CP=1/0.2=5
Enthalpy
The CP table Identification of the essential matches in the pinch region can be clarified using what we call CP table. In a CP table as will be shown in the graphs below, the CP values of the hot and the cold streams at the pinch are listed in descending order. Graph (A) below shows the grid diagram with a CP table for the HEN design/streams matching above the pinch at the pinch. Cold utility must not be used above the pinch, which means that hot streams must be cooled to pinch temperature by heat recovery. Hot utility can be used if necessary on the cold streams above the pinch. Thus it is essential to match hot streams above the pinch with a cold partner. In addition, if the hot stream is at pinch conditions, the cold stream it is to be matched with must also be at pinch conditions; otherwise the Tmin constraint will be violated. The graph (A) below shows a feasible match above the pinch that does not violate the Tmin.
Page 131 of 208
It is important to note here that the CP inequality constraint applies only when a match is made between two streams that are both at the pinch. Away from the pinch, temperature differences increase and it is no longer essential to obey the CP inequalities.
(A) Above The Pinch CPH <= CPC 0.25 0.15 0.30 0.20
Pinch
150 C
CP (MW/C)
2 4
250 C 200 C 180 C 230 C
150 C 150 C 140 C 140 C
0.15 0.25
1
0.20 0.30
3 140 C
The graph (B) below shows the grid diagram with a CP table for design below the pinch. Hot utility must not be used below the pinch, which means that cold streams must be heated to pinch temperature by heat recovery. Cold utility can be used, if necessary, on the hot streams below the pinch. Thus it is essential to match cold streams below the pinch with a cold partner.
Page 132 of 208
In addition, if the cold stream is at pinch conditions, the hot stream it is to be matched with also must be at pinch conditions; otherwise, the Tmin constraint will be violated. The graph (B) below shows a feasible match below the pinch that does not violate the Tmin. It is important to note here that the CP inequality constraint applies only when a match is made between two streams that are both at the pinch. Away from the pinch, temperature differences increase and it is no longer essential to obey the CP inequalities.
(B) Below The Pinch CPH >= CPC 0.25 0.15 0.20
Pinch 150 C
CP (MW/C)
150 C 150 C 140 C
40 C 80 C 20 C
0.15 0.25 0.20
140 C
Page 133 of 208
In conclusion there are some essential matches at the pinch or the region of minimum choice (ROMC) that need to be made around the pinch or the ROMC. The next task is to design a network that exhibit minimum number of units. In other words we need to decide how big should be the matches to minimize the HEN number of units. The tick-off heuristic Once the matches around the pinch have been chosen to satisfy the criteria for minimum energy, the design should be continued in such a manner as to keep capital cost to a minimum. One important criterion in the capital cost is the number of units. (Ofcourse, there is others that need to be addressed) Keeping the number of units at a minimum can be achieved using the tick-off heuristic. To tick off a stream, individual units are made as large as possible. In other words the smaller of the two heat duties on the streams being matched shall be taken completely. Graph (a) below shows the matches at the pinch with their duties maximized to tick-off streams. Stream #1 has been ticked off and stream #4 too. The heat exchangers duties are shown below. The heat exchanger that has the stream # 1 matched with part of stream #2 has a heat duty of 8 MW. While the heat exchanger that has the match between stream #4 and part of stream #3 has a heat duty of 12.5 MW. The design in graph (a) below can now be completed by satisfying the heating and cooling duties away from the pinch.
Page 134 of 208
(a) Matching Streams Above The Pinch CPH <= CPC 0.25 0.15 0.30 0.20
Pinch
150 C
CP (MW/C)
2 4
250 C 200 C 180 C
203.3
150 C 150 C 140 C
0.15 0.25
1
0.20 0.30
8 MW
230 C
187.7
140 C
12.5 MW
140 C
Cooling water must not be used above the pinch, therefore if there are hot streams above the pinch which the duties are not specified by pinch matches, additional process-to-process matches for more heat recovery shall be explored. In the graph (b) below the remaining part of the hot stream (stream # 2) is matched with stream #3 in another heat exchanger with heat duty of 7 MW to satisfy the residual cooling requirement for the hot streams above the pinch. Again the duty on the unit is maximized.
Page 135 of 208
(b) Matching Streams Above The Pinch CPH <= CPC 0.25 0.15 0.30 0.20
Pinch
150 C
CP (MW/C)
250 C
2
203.3
150 C 150 C 140 C
0.15 0.25
1
200 C
4
180 C 230 C
8 MW
205 187.7
0.20 0.30
140 C
12.5 MW
7 MW
140 C
Page 136 of 208
Finally, above the pinch the residual heating duty requirement of the cold streams must be satisfied. Since there are no hot streams left above the pinch and the cold stream # 3 needs to be heated up to 230 C, hot utility must be used as shown in graph ( c ) below. The heater duty is shown in the graph. The heater duty is the minimum heating duty calculated in the targeting phase which is equal to 7.5 MW.
(c) Matching Streams Above The Pinch CPH <= CPC 0.25 0.15 0.30 0.20
Pinch
150 C
CP (MW/C)
250 C
2
203.3
150 C 150 C 140 C
0.15 0.25
1
200 C
4
180 C 230 C
8 MW
205 187.7
0.20 0.30
140 C
12.5 MW
7 MW 7.5 MW
140 C
Page 137 of 208
Now let us consider the second half of the problem below the pinch, as shown in graph (a) below. The graph shows the pinch design with the streams ticked off. If there are any cold streams below the pinch for which the duties are not satisfied by the pinch matches, additional process-to-process heat recovery must be used, since hot utility must not be used. The graph shows that the hot stream # 4 has been ticked off and the heat exchanger used for the match has the highest possible duty of 17.5 MW.
(a) Matching Streams Below The Pinch CPH >= CPC 0.25 0.15 0.20
Pinch 150 C
CP (MW/C)
150 C 150 C 140 C

52.5
40 C 80 C 20 C
1
0.15 0.25 0.20
17.5 MW
140 C
Page 138 of 208
The hot stream#2 still needs to be cooled to 40C. This can be done via matching the stream with the cold stream # 1 as shown in graph (b) below.
(b) Matching Streams Below The Pinch CPH >= CPC 0.25 0.15 0.20
Pinch 150 C
CP (MW/C)
150 C 150 C 140 C

52.5
106.7
40 C 80 C 20 C
1
0.15 0.25 0.20
17.5 MW
6.5 MW
140 C
In graph (b) above we see additional match to satisfy the residual heating of the cold streams below the pinch. Again the duty load on the unit is maximized. The new match has a heat exchanger duty of 6.5 MW.
Page 139 of 208
Finally, below the pinch the residual cooling duty in the hot streams must be satisfied. Since there are no cold streams left below the pinch, cold utility must be used as shown in graph (c) below.
(c) Matching Streams Below The Pinch CPH >= CPC 0.25 0.15 0.20
Pinch 150 C
CP (MW/C)
150 C 150 C 140 C

52.5
106.7
40 C
10 MW
0.15 0.25 0.20
80 C 20 C
1 6.5 MW
17.5 MW
140 C
The graph (c) above shows that the remaining hot duties of the hot stream #2 that need to be removed from the stream to allow the hot stream temperature of 106.7 C reach its target temperature of 40 C is going to be handled using cold utility. The duty of the cold utility heat exchanger as shown in the graph (c) above is 10.0 MW. This duty is the minimum cooling utility that has been calculated during the targeting phase of the HEN design. Now the HEN design has been completed after finishing the below pinch design.
Page 140 of 208
The final design shown in graph below is just an augmentation of the above pinch design with below pinch design. The duty of the hot utility is 7.5 MW. It agrees with the QHmin. The duty on the cold utility is 10 MW. It also agrees with the QCmin. Both heating and cooling utilities agree with the QHmin and QCmin predicted by the composite curves and the problem table algorithm.
The Completed Design for the HEN problem

Pinch Pinch 150C 150 C
250 C
2 4
203.3
150 C 150 C 140 C
150 C 150 C 140 C

52.5
106.7
40 C
10 MW
200 C 180 C 230 C

8 MW
205 187.7
4 1
80 C
1 6.5 MW
17.5 MW
140 C
12.5 MW
7 MW 7.5 MW
140 C 140 C
Page 141 of 208
This design procedure is known as the pinch design method and it can be summarized as follows:
Divide the problem at the pinch into two separate problems The design for the separate problems is started at the region of minimum choice known as the pinch and then moving away Temperature feasibility requires constraints on the CP values to be satisfied for matches between streams at the pinch for the two problems above the pinch and below the pinch The loads on individual heat exchangers are determined using the tick-off heuristic to minimize the number of units Away from the pinch there is usually more freedom in the choice of matches. In this case the designer can choose on the basis of his/her process knowledge
Before we close the discussion about HEN design for new projects we need to mention at a glance the design needs for streams splitting. Streams splitting Stream splitting is sometimes necessary to overcome the CP constraints mentioned above and/or to avoid using cold utility above the pinch or hot utility below the pinch. Cooling utilities should not be used above the pinch. It means that all hot streams must be cooled to pinch temperature by heat recovery. If we have number of hot streams greater than the number of cold streams (Three hot streams and two cold streams) a problem will then arise. Since regardless of the CP values of the streams, there will be one of the hot streams that will not to be cooled to pinch temperature without some violation of the Tmin constraint. This problem can only be resolved by splitting a cold stream into two parallel branches. Thus in addition to the CP criterion, there is a stream number criterion above the pinch such that Sh<=Sc (above the pinch)
Where, Sh = number of hot streams at the pinch (including branches) Sc = number of cold streams at the pinch (including branches) If there had been more cold streams than hot streams in the design above the pinch, this would not have created a problem, since hot utility can be used above the pinch.
Page 142 of 208
Let us now consider the part of the design below the pinch. Here hot utility must not be used below the pinch. Thats mean that all cold streams must be heated to pinch temperature by heat recovery. Again if we have below the pinch the number of cold streams is greater than the number of hot streams (3 cold streams and two hot streams) regardless of the CP values one of the cold streams can not be heated to pinch temperature without some violation of the Tmin constraint. The problem can be solved by splitting a hot stream into two parallel branches. In such case each cold stream will have a partner with which to match and capable of heating it to pinch temperature. Thus there is also a stream number criterion below the pinch such that Sh<= Sc (below pinch) If we have more hot streams than cold streams below the pinch, this would not be a problem, since cold utility can be used below the pinch. It is instructive to mention here that, it is not only the stream number that creates the need to split streams at the pinch. Sometimes the CP inequality criteria for the streams above the pinch at the pinch and below the pinch at the pinch can not be met at the pinch without a stream split. It is important to emphasize here on the need to satisfy both criteria the stream population and the CP inequality. Number of hot streams above the pinch at the pinch needs to be greater than or equal the number of cold streams above the pinch at the pinch. If this is not the case then we need to split a hot stream to achieve this guideline. In the same time the CP of the hot stream above the pinch at the pinch shall be less than or equal to the CP of the cold stream above the pinch at the pinch in order to be able to match them in a heat exchanger. If this is not the case the cold stream need to be split into two. On the other hand, Number of cold streams below the pinch at the pinch needs to be smaller than or equal the number of hot streams above the pinch at the pinch. If this is not the case then we need to split a hot stream to achieve this guideline. In the same time the CP of the hot stream below the pinch at the pinch shall be greater than or equal to the CP of the cold stream below the pinch at the pinch in order to be able to match them in a heat exchanger. If this is not the case the cold stream need to be split into two. The graphs below present the flowchart that can be used for the overall approach.
Page 143 of 208
Stream-Splitting Above the Pinch

Stream data at Pinch
Yes
CPh<=CPc For pinch matches

NO
Yes
Sh<=Sc
NO
Split cold stream
Split hot stream
Place Matches
Page 144 of 208
Stream-Splitting Below the Pinch

Stream data at Pinch
Yes
CPh>=CPc For pinch matches

NO
Yes
Sh>=Sc
NO
Split cold stream
Split hot stream
Place Matches
HEN Retrofit using Pinch Design Method It is important to note here that for HEN projects retrofit where there is no significant integration already built in the plant the above method known as pinch design method (PDM ) is utilized with maximum re-use of existing exchangers. This method in retrofit applications will proceed as follows: a- Delete the existing network b- Re-design the HEN network by following the Pinch Design Method mentioned above c- Re-use existing exchangers in place of new ones between the same streams
Page 145 of 208
In situations where the existing network already involves many process to process heat exchangers, it is not appropriate to delete the entire network in order to apply the Pinch design Method. Instead, it is better to apply a method that makes incremental changes to the existing network, with a corresponding quantification of the benefits. This is particularly true of processes with diverging composite curves and significant use of intermediate utilities. Indeed, for such a process it is usually possible to discover a series of independent retrofit projects, each involving just a few modifications to the network. It is then possible to rank these projects and just choose the best ones according to some practical or economic criteria. Another important point regarding the stream splitting is that If a hot stream is need to be split into two parts with CP of 10 and 5 to satisfy the CP inequality criterion, we are doing here a subjective decision since the split can be anything that satisfies the CP criterion. In other words it can be 9 and 6 or 11 and 4 and so on. That is mean the stream splitting has several degrees of freedom in the branching flowrate that while can achieve the CP inequality criterion it can also used for optimizing the system. For instance by fixing the heat duty on the two heat exchangers for the split stream and changing the branch flowrate the temperature differences across each unit are changed. The best choice of-course can be done via optimization where the sizing and costing of the various units will be completed for the network with different branch flowrate. Later extensions of this method enabled the engineers to also consider investment cost during design, in particular the effect of each match on total heat transfer area. The Driving Force Plot makes is possible to evaluate graphically whether a suggested match is using reasonable driving forces compared with what is available in that temperature region of the process. The Remaining Problem Analysis is more quantitative tool, that provides figures for energy (E), number of units (U), heat transfer area (A) and total annual cost (TAC), if a suggested match is accepted. Adding actual figures for partial designs under development to target values for the remaining problem provides accumulated figures for TAC. Graph below shows an initial heat exchanger network for the process example studied in this chapter when the distillation column is not integrated with the rest of the process.
Page 146 of 208
mCp (kW/C)
Process Pinch
180 C
H1= 18 H2= 22 C1= 20 C2= 50
270 C
H1
220 C
235.6 C 3 2
180 C
Ca 360 kW 4 270 C Cb 440 kW
160 C
H2
210 C 2 1000 kW 210 C 190 C 177.6 C
180 C 1 160 C
60 C 50 C
4 2200 kW 160 C 1 880 kW
C1
H 1000 kW
3 620 kW
C2
160 C
The targeting phase concluded that (in this particular case) integrating the column with the rest of the process would not result in any energy savings, since the column operates across the process Pinch. Also notice that only HP steam and cooling water is used for external heating and cooling. The network in figure above has been established using the Pinch Design Method and is drawn using the Stream Grid. The initial heat exchanger network reaches the targets values for energy consumption (1000 kW of heating and 800 kW of cooling) and minimum number of units. For the case with two utilities, four process streams and strict decomposition at the process Pinch, the minimum number of units is (5-1) above Pinch and (4-1) below Pinch, in total 7 heat exchangers including heaters and coolers. It is important to notice that several initial networks may be generated. The Pinch Design Method provides rules for matching streams that eliminate certain configurations but still open up for alternatives. The larger the industrial problem is, the more alternatives exist, and the engineer is free to make choices based on practical considerations such as safety, operability, controllability, etc.
Page 147 of 208
In the small process example used in this chapter, only one significant alternative exists above Pinch. By splitting stream C2, it is possible to reduce the number of heat exchangers by one, as shown in graph below. While this example illustrates the existence of sub-networks above Pinch, stream splitting is more often used to be able to reach minimum energy consumption. The best example is crude preheat trains in oil refineries, where there is one large cold stream (the crude oil) and many hot streams (intermediate products and pumparounds from the distillation tower), and the crude is typically splitted in two, three or four branches before and after the desalter. Splitting of streams is also introduced to save total heat transfer area (better utilization of the available temperature driving forces), and in some rare cases splitting is also used to reduce the number of units, as indicated in graph below. Here, the heating needed by stream C2 matches exactly the cooling required for streams H1 and H2 above Pinch. An alternative initial Heat Exchanger Network for the process example
mCp (kW/C)
Process Pinch
180 C
H1= 18 H2= 22 C1= 20 C2= 50
270 C
H1
220 C
180 C 2 180 C 1 160 C 4 2200 kW 2 222.3 C 1620 kW 196.7 C 1 880 kW 26 24 160 C 50 160 C 4 270 C
Ca 360 kW Cb 440 kW
160 C
H2
210 C H 1000 kW 210 C
60 C 50 C
C1
C2
Page 148 of 208
Below Pinch there is also a design alternative, since the cooler could have been replaced by a match between H1 and C1, with a corresponding increase in the duty of cooler Cb and decrease in the duty for exchanger 4. Basic Concepts for Heat Recovery in HEN Retrofit While the majority of early days methods developed within Process Integration were related to the design of new plants, most of the projects in industry are trying to make the most out of existing facilities. Typically, these projects are related to improved operation, removal of plant bottlenecks, improved efficiency with respect to energy and raw material utilization, and the introduction of new technology into an existing process. Many terms are used for plant modifications, such as retrofit, revamp and debottlenecking. In this section, the term retrofit is used for projects trying to reduce energy consumption in the most economic way. Typical economic parameters or constraints are maximum allowed values for Payback Time and Investment Cost. The objective of a retrofit project is then to save as much energy as possible while satisfying these economic constraints. The economy of most energy saving projects (cost of new equipment versus reductions in operating cost) is not good enough to include the losses in production if the plant has to be stopped for a period of time while the modifications are installed. Thus, the timing of retrofit projects into regular plant maintenance periods is extremely important. Further, the best retrofit projects are the ones that combine pure energy saving features with more general plant modifications. Some Useful Representations Grassroots Pinch Analysis can and has been used to a large extent in industry to establish the potential for energy savings in existing plants. When comparing the current energy consumption with grassroots targets, however, the identified potentials tend to be rather optimistic. In the process industries there is no "second hand" market, thus one of the prime objectives in retrofit projects is to try to improve the utilization of already invested and installed equipment. There will be discrepancies in the existing design that cannot be completely removed, only improved by smaller or larger process modifications. As a result, the optimal heat exchanger network after retrofit is likely to be quite different from the optimal grassroots design.
Page 149 of 208
Area-Energy Plot for Heat Exchanger Networks
Area
Aexist
c
Amin
Emin
Eexist
Energy
In an existing plant, the heat recovery system can suffer from two types of errors, as illustrated in graph. Each point on the curved line indicates the minimum amount of heat transfer area that is required to have a certain energy consumption (or level of heat recovery). Similarly, the curve also indicates minimum energy consumption for a given total heat transfer area. The points (a) to (e) represent different design solutions that will be discussed in the following. The curved line in graph, also referred to as the AreaEnergy Plot, is constructed by calculating minimum target values for energy and area as indicated before for different values of the minimum allowable approach temperature, Tmin. With small values of Tmin, the minimum area target is large, while the energy target is low, and opposite for large values of Tmin.
Page 150 of 208
Assume that design (c) is the optimal grassroots heat exchanger network, with an optimal trade-off between operating cost and investment cost for the current energy and area prices. Network (a) has been correctly designed in the sense that it uses minimum area to achieve a certain level of heat recovery. Most likely, this design has been established by the Pinch Design Method. The trade-off in this design is wrong, however, as it uses more energy than would have been optimal with the current prices. In a retrofit project, it will be very hard and costly to improve this network. Moving along the curve from (a) to (c) would mean that a number of heat exchangers would have to be taken out of the network. What would be done in practice is to keep most of the existing exchangers and invest in some new ones. The corresponding retrofit project would move along a curve above the minimum target line, and this curve would be steeper than the target line. Next, consider design (e), which is located far above the target line. If this had been a suggested new design, both investments cost (area) and operating cost (energy) could have been reduced as indicated by A and E. If this is an existing network, however, it is not economically tractable to try to reach design (c), since that would involve throwing away a large number of invested heat exchangers. Again, the retrofit project would follow a curve to the left, but in this case it would be flat in the beginning, since the existing network has major errors that can be corrected by moderate investments, such as re-piping and the addition of strategically placed new heat exchangers. After correcting the most obvious errors in the existing design, the cost of recovering additional heat will gradually become more costly. This means that the retrofit curve would become steeper, and payback time therefore increases with the amount of energy saved. While network (a) is a "good" design (unfortunately with a wrong tradeoff), network (e) is a "poor" design, since it uses much more energy than what could have been achieved with the amount of invested heat transfer area. The errors in design (e) are important in retrofit projects and will be discussed in detail in this section. It should also be mentioned that the minimum area figures used to establish the target curve in figure 6.1 actually require a large number of heat exchangers, splitters and mixers (referred to as the Spaghetti Design). Thus, one would never design on the target line, but some small distance above. The figure above is a quantitative tool to identify the potential for improved heat recovery, and at the same time a qualitative picture of the situation indicating how costly the corresponding retrofit projects will be. What are needed next are some guidelines on how to actually modify the network. The reason why an existing design, such as network (e) is using more than the minimum amount of energy (both heating and cooling), is the fact that heat is being transferred across the heat recovery Pinch. Such heat transfer can take three different forms as will be mentioned later.
Page 151 of 208
The heat recovery Pinch divides the process into a heat deficit part above Pinch and a heat surplus part below Pinch. Of course, it would not make sense to transfer heat from the deficit part to the surplus part. Nevertheless, when heat exchanger networks are designed without the knowledge about the heat recovery Pinch, such heat transfer is often inevitable. This is why large potentials for energy savings have been identified in existing plants, and the more complex these processes are, the more likely it is that considerable cross Pinch heat transfer takes place. Typical examples are petrochemical plants and oil refineries, however, significant potentials have also been identified in other industries as well. There may also be practical reasons for such heat transfer across the Pinch. One of the major limitations of the Composite Curves and the corresponding Pinch analysis, is that hot and cold streams are regarded to be heating and cooling resources that can be used without limitation. In practice, however, there will be match combinations among hot and cold streams that one would avoid. Examples include safety considerations, geographical distance, start-up considerations, ensuring product purity, etc. In many of these cases, heat transfer across the Pinch is inevitable, however, there are some degrees of freedom in how this heat transfer takes place. Heat can be transferred across the heat recovery Pinch in the following three ways: Heat transfer from a hot stream above Pinch to a cold stream below Pinch: QPp Heating a cold stream below Pinch with hot utility, such as steam: QPh Cooling a hot stream above Pinch with cold utility, such as cooling water: QPc The total Energy Penalty (QP) for heat transfer across the Pinch is then the sum of these individual heat flow components: QP = QPp + QPh + QPc This penalty is then the difference between the current energy consumption and the minimum energy consumption for a given value of Tmin: QH,exist = QH,min + QP QC,exist = QC,min + QP
Page 152 of 208
The three components of penalty heat flow (QPp, QPh and QPc) can be considered as variables that can be used to take advantage of the situation when practical constraints result in an energy penalty. When trying to minimize the cost penalty of such constraints, the three variables provide two degrees of freedom. The following advantages can be taken from a constrained situation: QPp means heat transfer at larger driving forces, thus heat transfer area is reduced. QPc can be realized as steam production, if the Pinch temperature is high enough. QPh means that a cheaper hot utility with lower temperature can be used. Since the energy target depends on the chosen value of Tmin, the corresponding potential for reduced energy consumption is larger for a smaller value of Tmin. The corresponding retrofit project will, however, also be more complex and costly. While targeting methods exist for the retrofit case that can identify a proper value for Tmin (will be described later), it is common practice in industry to use a larger value for Tmin in a retrofit situation than the corresponding and optimal value of Tmin in a grassroots case. Having explained the features of an existing design that is responsible for a larger than minimum energy consumption, the next logical step is to look at the actual heat exchanger network to identify which process/process heat exchangers, external heaters and coolers that are actually transferring heat across pinch. The Stream Grid is an excellent tool for this purpose. In the following graph, an existing heat exchanger network is drawn in a stream grid in such a way that the relative position (in temperature) to the Pinch is indicated for all units.
Page 153 of 208
Cross Pinch Heat Transfer in a Stream Grid

mCp (kW/C)
Process Pinch
180 C
H1= 18 H2= 22 C1= 20 C2= 50
270 C
H1
220 C
214.4 C
Ca
180 C 980 kW
160 C
H2
210 C 1 1000 kW 210 C H 2500 kW 160 C
120 C Cb 1320 kW 4 2200 kW 160 C
60 C 50 C
C1
C2
160 C
A Preliminary Retrofit Discussion It is now easy to identify which heat exchangers that transfer heat across the Pinch, and what amount of heat that is transferred across Pinch in each of these units. The sum of all these cross Pinch occurrences should add up to the total energy penalty. It should be noted, however, that there sometimes are cases where a heat exchanger operating with small temperature driving forces is transferring heat from below to above Pinch. These heat flows must then be subtracted when calculating the total energy penalty. The heat exchanger network in the graph above uses 2500 kW of hot utility and 2300 kW of cold utility. The corresponding minimum target values for Tmin = 20C are QH,min = 1000 kW and QC,min = 800 kW (The total energy penalty for this existing design is thus): QP = QH,exist - QH,min = QC,exist - QC,min = 1500 kW
Page 154 of 208
For the network in this graph, cross Pinch heat transfer can be identified in heat exchanger (2) and cooler (Ca). The actual amount of heat transfer across the Pinch in these units can be calculated as follows: Exchanger (2): QPP = 22 (220 - 180) = 880 kW Cooler (Ca): QPC = 18 (214.4 - 180) = 620 kW In this case, there is no external heating below Pinch, and the total energy penalty can be calculated from the occurrences of cross Pinch heat transfer: QP = QPP + QPH + QPC = 880 + 620 + 0 = 1500 kW A Retrofit Solution reaching Minimum Energy Consumption is shown in the graph below.
mCp (kW/C)
Process Pinch
180 C
H1= 18 H2= 22 C1= 20 C2= 50
270 C
H1
220 C
235.6 C 3 1
180 C
Ca 360 kW 2 80 C Cb 440 kW 2 2200 kW
160 C
H2
210 C 1 1000 kW 210 C 190 C 177.6 C
180 C 4 160 C
60 C 50 C
C1
160 C 4 880 kW
H 1000 kW
3 620 kW
C2
160 C
Page 155 of 208
Obvious retrofit projects should involve attempts to reduce heat transfer across the Pinch. Hot stream H2 is a heating resource above Pinch that could be used to heat up cold stream C2 and thus reduce the use of steam in the heater (H). Hot stream H1 is also a heating resource above Pinch, where some heat in the existing design is lost to cooling water. Trying to realize the total potential for energy savings (1500 kW), would involve two new heat exchangers (3 and 4) and additional area in the existing ones (1 and 2), due to reduced driving forces. The corresponding heat exchanger network shown in figure above is actually identical to the initial MER design for the grassroots case shown before. Without actually performing cost calculations, it is obvious that the retrofitted network in graph above will be very expensive. It is almost an entirely new heat exchanger network. An alternative solution would be to try to recover some of the heat that is lost from hot stream H1 into cooling water, by adding a new unit between hot stream H1 and cold stream C2. In this case, the existing heat exchangers are not modified (no additional area is needed), and the simple question is whether the saving of 620 kW of steam and cooling water will justify the investment in a new heat exchanger (3). The corresponding network is shown in the figure below. A cheaper Retrofit Solution recovering part of the Potential Energy Savings
mCp (kW/C)
Process Pinch
180 C
H1= 18 H2= 22 C1= 20 C2= 50
270 C
H1
220 C
214.4 C
180 C
Ca 360 kW 120 C Cb 1320 kW 2 2200 kW
160 C
H2
210 C 1 1000 kW 210 C H 1880 kW 172.4 C 160 C
60 C 50 C
C1
160 C 3 620 kW 160 C
C2
Page 156 of 208
Having shown some of the useful representations and indicated a possible "thinking" in retrofit situations based on cross Pinch heat transfer, the remaining part of this chapter will be devoted to briefly presenting the methods that can be used for heat exchanger network retrofit. Similar to the grassroots case, there are four distinct phases also for retrofit design: 1) Data Extraction 2) Targeting 3) Design 4) Optimization/improvement There will, however, be significant differences in all of these phases when compared to the grassroots situation. These differences and the new objectives will be highlighted in the description of each of these phases. Data Extraction in Retrofit Applications While there are a number of similarities between data extraction in the retrofit situation and the grassroots case there are also significant differences that will be highlighted here. In both cases, data extraction is a time consuming and critical activity for the outcome of a Process Integration project. Typically, for a new design there will be material and energy balances available either manually derived or based on a simulation model. A rigorous simulation model has the distinct advantage that stream data can be extracted directly and even automatically with todays software. Unfortunately, such models may not always be available for an existing plant. In general, the following are possible sources for data that are needed in a retrofit heat recovery project, and often these sources have to be used in combination: Measurements (that are often not complete and not reliable) Design data (that are often outdated after plant modifications) Simulation models (that may not always reflect true plant behavior)
Page 157 of 208
As a result, data reconciliation is important in retrofit projects. If measurements indicate that heat extracted from a hot stream in an existing heat exchanger does not match the heat absorbed by the cold stream in the same exchanger, it is necessary to analyze the situation. Stream data must be modified in such a way that heat balance is obtained; otherwise the heat recovery project will produce unrealistic results. It is important to notice that data accuracy is most important in the near Pinch region of the plant. Thus, it is common practice to try to establish a first draft of the Composite Curves, and then try to improve the accuracy only for process streams in the near Pinch region. Another typical retrofit issue is related to which streams to include in the analysis. There may be a number of practical considerations suggesting that certain streams should not be included, since heat integration of these streams could cause operational problems. It is, however, good practice to start by including all streams that need heating or cooling, and then later exclude these streams one by one from the analysis. In this way, the engineer will know the loss in heat recovery potential from excluding certain streams. In retrofit projects it is not necessary to iterate between data extraction and targeting, since the basic process (reactors, separators) is given and cannot easily be modified for improved heat recovery potential as the case is for grassroots projects. It would also be expensive to modify these process units, and would seldom pay off in pure energy based projects. Retrofit Targets Targeting in the Retrofit situation is far more difficult than Grassroots design because a number of different changes can be made to the heat exchanger network in order to reduce energy consumption. Typically, these modifications include: Addition of a new heat exchanger Additional area to an existing unit (for example a new shell) Change internals in heat exchangers Modify piping on one side of the exchanger Modify piping on both sides of the exchanger Moving a heat exchanger to a new location
Page 158 of 208
Most of these retrofit actions will change the operating conditions for many of the heat exchangers, and a rigorous rating exercise is required to evaluate whether an existing unit will be able to operate in the new situation. The cost function for the retrofit project will exhibit a discontinuity whenever a heat exchanger switches from being large enough to become too small for the new operation. In other words, the targeting of capital investment (new heat transfer area and new units) is much more difficult than in the grassroots case. Energy consumption, on the other hand, is much easier to predict, however, knowing the savings in energy cost is of limited value if it is not correctly linked with its corresponding investment cost. That is the true challenge in retrofit targeting. Different T Representations The ultimate goal of the targeting exercise is to establish a good starting value for the level of heat recovery. In grassroots heat exchanger network design, the parameter Tmin (minimum approach temperature) was used to represent this level of heat recovery. In most industrial processes, it does not make sense to require that all heat exchangers (and thus all process streams and utilities) obey the same minimum value for driving forces, since streams (and utilities) in general have very different heat transfer coefficients. Quite often, the difference in film heat transfer coefficients can be two orders of magnitude. Thus, some heat exchangers require large T-values in order to avoid excessive heat transfer area, while other units manage well with much smaller T-values. Note: The effect of the T-values does not only get reflected in the capital but also in the level of energy consumption. We should at least acknowledge the need for two different approach temperatures: HRAT = Heat Recovery Approach Temperature EMAT = Exchanger Minimum Approach Temperature While HRAT, as the name indicates, is a key parameter for the level of heat recovery (it is simply defined as the smallest vertical distance between the Composite Curves), EMAT is the minimum allowable temperature difference for the individual heat exchangers. In order to reach a certain level of heat recovery, (as specified by HRAT), the following inequality must be satisfied:
0.0 EMAT HRAT

Page 159 of 208
it is also possible to assign individual contributions to the minimum driving forces for each stream and utility. Typically, these T contributions should reflect heat transfer conditions, but they can also be used to represent the need for expensive material of construction, expensive heat exchanger types, etc. In this case, EMAT becomes stream dependent, and the following must be satisfied for a match between hot stream/utility (i) and cold stream/utility (j): Ti,j >= EMAT (i,j) = Ti + Tj (where Ti and Tj are the individual stream contributions) To illustrate how these different T-values apply in retrofit situations, consider a typical oil refinery with a crude preheat train that warms up the crude from ambient temperature to the inlet of the furnace just before the crude fractionation tower. This is a complex heat exchanger network with many units, a large number of stream splits and considerable heat recovery from various hot streams in the refinery. In retrofit projects for such plants, it is common to design for a level of heat recovery that corresponds to HRAT = 30C, however, the actual energy consumption in many such crude preheat trains corresponds to a value of HRAT well above 50C. At the same time, there will be some heat exchangers typically where T in one end of the units is in the range between 10 and 15C. Thus, EMAT and the individual contributions Ti and Tj are considerably less than HRAT. A Simple Energy Target An obvious way to establish a target for energy savings in a retrofit project is to calculate the minimum external heating requirements for different values of HRAT (previously referred to as Tmin). One of these values of HRAT (typically a large one) corresponds to the current energy consumption, and the targeting exercise then becomes the identification of a new value of HRAT that is less than the existing value of HRAT: HRATnew HRATexisting By plotting minimum energy consumption (or minimum energy cost in the case of multiple utilities) as a function of HRAT, it is possible to identify potential starting values of HRAT for the retrofit project.
Page 160 of 208
A typical Energy Target Plot for the Retrofit Case
Qh,min
Qh,exist
E
Qh,new
a b
c
HRAT-new HRAT-exist
HRAT
Consider this Energy Target Plot, where the change in the slope illustrates the typical situation that minimum energy consumption does not always increase directly proportional with minimum driving forces. As a result, there are certain levels of heat recovery (represented by HRAT) that are more likely to be good retrofit candidates than others. Consider point (a) in this figure when trying to move towards larger energy savings, the change in QH,min is relatively small, while the reduction in HRAT is considerable. Normally, this means large investments with moderate savings. The graph also indicates how the targeted savings in energy consumption (E) can be read from the diagram for different values of HRAT. Also, by looking at the graph, one may conclude that point (a) seems to save too little energy, while point (c) involves too large investments. Qualitatively, it may look as if point (b) provides a good trade-off between investments and savings in the retrofit case; thus HRATnew is a good starting value for the retrofit project. Cross Pinch Analysis will then be performed, where the existing heat exchanger network is drawn in a stream grid with a Pinch point according to; Tmin = HRATnew.
Page 161 of 208
Targets for Area and Investment Cost As stated before targeting for heat transfer area and investment cost is far more complicated and uncertain in the retrofit situation than in the grassroots case. The identification of promising starting points in work in some cases however, there is a need to quantify not only the energy saving part, but also the investment in new equipment and changes in piping. Within Pinch Analysis, a Retrofit Targeting procedure has been proposed that is based on the concept of Area Efficiency. This parameter can be easily obtained from the existing design and can be mathematically formulated as:
= [ Amin ] / [ Aexist ]
Where = Area Efficiency Amin = Minimum area for the current level of heat recovery (HRATexist) Aexist = Total heat transfer area in the existing network A conservative assumption is that any new invested heat transfer area will at least have the same utilization level (area efficiency) as the installed area. This assumption (also referred to as the constant- approach) proved to work nicely for oil refineries and crude preheat trains, where area efficiency in existing plants was quite high (above 80%), while it did not work equally well in other industries. In processes with less heat integration, the constant assumption can be too conservative. Attempts have been made to overcome this problem; one is the so-called incremental approach which means that area efficiency will change (improve) during the retrofit project.
Page 162 of 208
A typical Retrofit Target Plot for Savings and Investment Cost
Savings (US$/yr)
Pb=3 Pb=2 Pb=1
Inv-max
Investment (US$)
Irrespective of the actual approach chosen, using some kind of assumptions about area efficiency for new heat exchangers or additional shells, it becomes possible to estimate the need for new area (and thus investment cost) when targeting for different values of HRAT. By combining target values for area and energy for different values of HRAT (starting with the existing HRAT and then decreasing this value gradually), the Retrofit Target Curve shown in graph above can be obtained. The curved line in the graph shows corresponding values for annual savings in energy cost and the total investment for new equipment (including installation). Each point on the curve represents a certain level of heat recovery (HRAT), starting in the origin of the diagram, where there are no investments made and no savings earned. Thus, the origin represents the situation before the retrofit project is started, and moving along the curve to the right means heading for smaller HRAT values and higher levels of heat recovery. As discussed before, the most obvious errors in the existing network can be corrected first, often with small or moderate investments. Thus, the target line is initially steep, but then becomes more flat as we move towards higher levels of heat recovery.
Page 163 of 208
Payback Time is simply defined as the Investment Cost divided with the Annual Savings in Operating Cost (energy). Considering the nature of the target curve in graph below it is obvious that Payback Time increases as we move along the curve towards larger energy savings. The dashed lines in the diagram illustrate typically Payback Times (PB) of one, two and three years. It is also quite common that management has set an upper limit on the investment that will be put into a certain energy saving project (indicated as Invmax in graph. There will also be constraints on the Payback Time, and depending on whether maximum Payback Time for this particular example is set to two or three years, the retrofit targeting exercise will identify points (b) if minimum payback is two years or point (c) limited by maximum investment, if maximum payback is three years. Points (b) and (c) in graph correspond to different values of HRAT, which means that a target for the level of heat recovery has been identified. This target is an improvement compared with the more simplified discussion before, since investment cost has been included and quantified, even though there are large uncertainties in these numbers. Again, once the new value for HRAT has been identified, the next stage is a cross Pinch analysis. Retrofit Design The Cross Pinch Analysis mentioned in the previous section is a good starting point for the design exercise. The first methods suggested to remove heat exchangers that transferred heat across the Pinch and to try to reuse these units in new locations. Since, however, heat exchangers in most cases are tailor made for a certain application (flowrates and types of streams) it is not easy and quite expensive to follow this approach. Temperature Shifting of Heat Exchangers Instead it is suggested to shift heat exchangers away from a cross Pinch situation. This shifting involves changes in operating conditions for the unit in such a way that hot stream temperatures are reduced and/or cold stream temperatures increased. The result is that cross Pinch heat transfer in that particular unit is reduced and possibly eliminated. Heating resources are released above Pinch and or cooling resources are released below Pinch. Consider again the existing heat exchanger network studied earlier with cross pinch situation. The cooler Ca and heat exchanger (2) are transferring a total of 1500 kW across Pinch, which is why external heating (2500 kW) and cooling (2300 kW) requirements are larger than the established minimum figures (1000 kW of heating and 800 kW of cooling). The shifting procedure means that the inlet temperature of stream H1 to the cooler Ca should be reduced from 214.4C to at least 180C (Pinch temperature for hot streams). This will release a heating resource from stream H1 above Pinch equal to 620 kW, and the duty of cooler Ca is reduced from 980 kW to 360 kW.
Page 164 of 208
Similarly, the inlet temperature of stream H2 to heat exchanger (2) should be reduced from 220C to 180C. Assuming that the duty of this unit remains unchanged at 2200 kW (should always be questioned during network optimization), the duty of cooler Cb will be reduced by 880 kW to 440 kW. Figure below shows the incomplete network after these shifting operations. As indicated by the rectangles, there are two heating resources that have been released and not yet utilized above Pinch. Incomplete Heat Exchanger Network after Shifting
mCp (kW/C)
Process Pinch
180 C
H1= 18 H2= 22 C1= 20 C2= 50
270 C
H1
220 C
1 214.4 C 620 kW
180 C
Ca 360 kW 2 120 C Cb 440 kW
160 C
H2
880 kW 210 C 1 1000 kW 210 C H 2500 kW 160 C 160 C
60 C 50 C
2 2200 kW
C1
C2
160 C
Introduction of New Heat Exchangers The next obvious question is how to utilize these new heating resources above Pinch. Since cold stream C1 already is fully covered through heat recovery from hot stream H1, the obvious option is to try to use heat from hot streams H1 and H2 to partially heat cold stream C2 in order to reduce steam consumption in the heater.
Page 165 of 208
Following the basic philosophy of the Pinch Design Method, cold stream C2 cannot fully utilize the two new heating resources (would involve taking both streams H1 and H2 down to Pinch temperature) unless stream C2 is split into two branches. Since mCp for stream C2 (50 kW/C) is larger than the sum of mCp (18+22 kW/C) for streams H1 and H2, this is a feasible option. Alternatively, the heating resource related to hot stream H1 could be shifted to the beginning (hottest part) of the stream. This option has already been shown in the retrofit solution that reaches the minimum energy consumption. However, as indicated in this solution, heat exchanger (1) has considerably reduced driving forces and additional area is inevitable. The graph below shows the initial retrofitted heat exchanger network when the stream split option is chosen. In this case, the operating conditions duty and temperatures for heat exchanger (1) is unchanged, and no additional area is needed. Heat exchanger (2) has, however, reduced driving forces with the same duty, and additional area is needed as indicated. A comparison with the alternative retrofit design for maximum energy recovery will be made before going into the optimization stage. Complete Retrofitted Heat Exchanger Network with Stream Splitting
mCp (kW/C)
Process Pinch
180 C
H1= 18 H2= 22 C1= 20 C2= 50
270 C
H1
220 C
214.4 C 3
180 C
Ca 360 kW 2 80 C Cb 440 kW
160 C
H2
210 C 1 1000 kW 3 620 kW
180 C 4 160 C
60 C 50 C
2 2200 kW 50 4 880 kW 160 C 160 C
C1
210 C
190 C H 1000 kW
C2
Page 166 of 208
In order to compare the two alternative initial retrofit heat exchanger networks in the maximum energy recovery solution and the above graph, details about heat transfer conditions and cost equations are needed. In this particular case, we do have information about heat transfer coefficients for streams and utilities (table below), however, for the purpose of this manual we do not want to go into detailed cost calculations. Instead, comparison between the two alternatives will be made on the basis of a simple UA analysis. It is easy from the heat transfer equation to calculate UA-values for the heat exchangers before and after retrofit modifications. If we assume that the units are pure counter current, UA-values can be obtained from: UA = Q / TLM The table below shows UA-values for existing and new heat exchangers before and after retrofit modifications for the two alternative designs A and B. Utility exchangers are not included, since the duty of these units is reduced in such a way that no additional area is needed (actually, these units will not be fully utilized after the retrofit modifications). Isothermal mixing is assumed for stream C2 after the split. As indicated in the table below the UA analysis does not give any strong preference for design A or design B. The difference in total UA needed in the retrofitted networks is not significant, and a stronger argument for choosing design B is probably that the number of modifications is less, since there is no change needed for heat exchanger (1) in this case, however, there is a stream split introduced. Heat Exchanger 1 2 3 4 Total Existing Design 17.49 33.91 0 0 51.4 Retrofit A 44.12 89.2 9.07 29.52 171.91 Retrofit B 17.49 89.2 27.99 35.68 170.36
While both design A and design B fully recover the energy saving potential of 1500 kW, in most cases one can only justify economically to realize some fraction of this potential. The figures for UA listed in table indicate that heat transfer area must be more than tripled in order to reduce energy consumption to its minimum for HRAT = 20C. Thus, more recent retrofit methods use a greedy approach trying to identify the most economic retrofit projects with the fewest number of topological changes.
Page 167 of 208
Matrix Methods for Retrofit Design Some interesting matrix based methods have also been proposed for heat exchanger network retrofit situations. It focused on heat transfer area in a method where targeting and design are closely linked. The so-called Area Matrix method is an adoption of the vertical heat transfer model. For various levels of heat recovery, the best vertical match area contribution is found using Linear Programming (LP). The result is a significantly improved retrofit area targeting method when compared with the constant or incremental methods mentioned before. While the Area Matrix method primarily is a targeting procedure, the results from the LP optimization can also be used for retrofit design. Another matrix based method for retrofit design is the Cost Matrix method developed. The method is based on the experience from a number of retrofit projects that other costs such as pumping and piping may have a larger influence on the optimal design than the number of units and heat transfer area. A Cost Matrix for possible matches is established, where the cost for each match is estimated taking into account parameters such as physical distance between process streams, material requirements, type of heat exchangers, auxiliary equipment (such as valves), heat transfer coefficients, space requirements, pumping cost, maintenance cost and fouling. The method uses the greedy approach due to its sequential nature, and there is no targeting involved. More Recent Retrofit Methods A number of more recent methods for retrofit heat exchanger networks using optimization (Mathematical Programming) to a large extent have been developed. Due to the rather complex nature of these methods, however, they are only briefly mentioned here in this course manual. The complexity of these methods also means that software is an absolute requirement. Typically, these methods acknowledge the fact that only a few carefully selected modifications will be economically worthwhile, and the approach is to identify these retrofit actions.
Page 168 of 208
2.5
Heat Exchangers Network (HEN) Improvement
Heat Exchanger Networks for maximum energy recovery established by the Pinch Design Method should only be regarded as initial designs. The strict decomposition at the Pinch normally results in networks with stream splits and a few rather small units. As mentioned above, the basic Pinch Design Method focuses on minimum energy consumption while using the fewest number of units. Even though extensions such as the Driving Force Plot and the Remaining Problem Analysis help the engineer to also minimize total heat transfer area, Total Annual Cost is not necessarily at its minimum, and some final optimization is required. With a good initial value for Tmin, only minor network changes (described as design evolutions by Linnhoff and Hindmarsh, 1983) are required in most cases. The matches of the initial network depend on the Pinch location, and since the Pinch point depends on the value of Tmin, this becomes a key parameter in Pinch based methods. The Degrees of Freedom available for network optimization are the following: Since the initial network is produced by respecting strict Pinch decomposition, there will be more than the minimum number of units. Thus, there are Heat Load Loops in the network, where 2 or 4 or 6, etc., heat exchangers (including heaters and coolers) have duties than can be modified in a systematic way without changing the stream target temperatures. Such loops can be used to remove small units in the network (discrete optimization) or simply to obtain an area distribution in the network with lower total annual cost (continuous optimization). There will also be Heat Load Paths from a hot utility exchanger through some of the process/process exchangers to a cold utility exchanger. These paths can be used to restore unacceptable temperature driving forces in some units after manipulation of heat load loops. Since increasing the duties of utility exchangers will affect the energy/area trade-off, this procedure has similarities to shifting the Composite Curves for the overall problem. A heat load path, however, affects only a limited number of units. In some cases, such heat load paths can also be used to remove small units. Flowrates of the individual branches of a Stream Split can be varied in order to reduce total heat transfer area (or actually investment cost) of the heat exchangers involved. This is a local optimization affecting a limited number of units, but interactions exist between this optimization and the manipulation of heat load loops and paths.
Page 169 of 208
The two HEN (heat exchanger networks) designs introduced above reflect maximum heat recovery (MER) solution. This solution may require more heat exchangers and more piping. The MER design also has temperature driving forces that are at least 20C in all its heat exchangers. Other designs may have different heat exchanger(s) with larger or smaller duties and may be only 10C temperature difference in the hot end in one or more of its streams. In these arguments, the distillation column and its use of HP steam and cooling water is not included, since the column is operated in the same way. In summary, network optimization can be carried out as a design evolution, without large modifications to the basic network structure. This method requires a good initial design, as the ones that can be established by the Pinch Design Method. In practice, cost information is required to actually optimize the network, but the basic strategy outlined here is still valid: a) Identify a good starting value for Tmin by pre-optimization based on individual targets for Energy, Area and Units (also referred to as SuperTargeting). b) Design an MER network using the Pinch Design Method. c) Remove the smallest unit by breaking a Heat Load Loop. d) Restore driving forces by manipulating a Heat Load Path. One of the major limitations in is the assumption of a global value of Tmin for all process streams and heat exchangers. In industrial applications, differences in heat transfer coefficients must be accounted for in Targeting, Design and Optimization. Another limitation is the fact that sequential procedures as the one outlined here have problems handling complicated multiple trade-offs and so-called topology traps.
Page 170 of 208
Network Improvement in Retrofit Applications As mentioned in the previous section, optimization is used in some of the more recent retrofit methods for network design, and the distinct classification into targeting, design and optimization has been reduced and almost eliminated. This part will discuss how initial retrofit designs developed using the methods described before can be improved economically and simplified with respect to network structure, using the same optimization philosophy as in the grassroots case. Degrees of freedom in the form of heat load loops and paths as well as stream splits can be used to improve the initial retrofit design. One important new aspect in the retrofit case is to maximize the utilization of existing heat exchangers. After the shifting of cross Pinch heat exchangers and the introduction of new units, some of the existing heat exchangers may have a reduced duty and therefore no longer require all the area installed. In such cases it may be worthwhile to shift duty in heat load loops and/or paths until the existing units are better utilized. Similar to grassroots situations, retrofit network optimization is a combination of discrete and continuous adjustments. The discrete part takes care of the removal of small new heat exchangers or small area additions to existing units, while the continuous part takes care of the trade-off between investment cost and obtained energy savings. The continuous part also includes, as mentioned above, the maximum utilization (if possible) of existing units. The cheaper retrofit solution that recover part of the potential energy savings introduced earlier in this retrofit part of the chapter shown again below shows an alternative retrofit heat exchanger network for the example problem, where only one topological modification is suggested.
Page 171 of 208
mCp (kW/C)
Process Pinch
180 C
H1= 18 H2= 22 C1= 20 C2= 50
270 C
H1
220 C
214.4 C
180 C
Ca 360 kW 120 C Cb 1320 kW 2 2200 kW
160 C
H2
210 C 1 1000 kW 210 C H 1880 kW 172.4 C 160 C
60 C 50 C
C1
160 C 3 620 kW 160 C
C2
The introduction of the new heat exchanger (3) between H1 and C2 recovers heat that is lost to cooling water above Pinch in the existing design. In this retrofit alternative, the existing heat exchangers (1) and (2) are not changed, and the optimization simply becomes a one-dimensional search to identify the largest duty for heat exchanger (3) that satisfies constraints such as maximum Payback Time and maximum Investment Cost.
Page 172 of 208
3.0
Pinch Technology and Mathematical Programming Hybrid Methods
Mathematical Programming is a class of methods for solving constrained Optimization problems. Since both continuous and discrete (or binary) variables can be used in the corresponding Mathematical Programming models, these methods are perfectly suited for typical design tasks encountered in Process Synthesis and Process Integration. The binary variables (0 or 1) can be used to model discrete decisions such as selection of equipment, the interconnection and sequencing of equipment and other non-continuous choices made by the designer. The continuous variables can be used to model process stream properties such as flowrate, temperature and pressure, as well as operating conditions and geometrical sizes such as area and volume of process equipment. While Mathematical Programming belongs to the so-called deterministic optimization methods, there have also been applications of stochastic or non-deterministic methods such as Simulated Annealing and Genetic Algorithms in Process Integration. These methods have been used to overcome numerical problems such as local optima caused by non-convexities in the models, discontinuities and combinatorial explosion. It is, however, beyond the scope of this course manual to describe these methods. Use of Simulated Annealing in Process Integration has been described in several papers and literatures. Even though Pinch Analysis is used routinely in a large number of industrial companies, indicating that the technology is realistic and applicable, there are a number of limitations inherent in Pinch based methods. These limitations are actually the major motivation for using Mathematical Programming in energy Integration. One of the most important and basic concepts in Pinch Analysis is the Composite Curves that in a single diagram gives the engineer a birds eyes view of the opportunities for heat recovery in the total process. The diagram provides details about the bottleneck(s) for heat recovery (Process and Utility Pinches), the minimum external heating and cooling requirements as a function of the specification of minimum allowed driving forces in the heat exchangers, and an indication of the total need for heat transfer area.
Page 173 of 208
It is, however, an underlying assumption that all hot and cold streams are resources that can be used without limitation for heat recovery purposes. Once the streams have been merged into the hot or cold composite curve, their identity is lost. If pairs of hot and cold streams are not allowed to exchange heat (for reasons such as safety, operability, piping difficulties, contamination prevention, etc.) it becomes extremely difficult to evaluate the effect on heat recovery of such forbidden matches from the composite curves. Using Mathematical Programming, however, it is extremely simple to formulate such situations, and the corresponding solution phase actually becomes easier. In Pinch Analysis, there is also a combination of Heuristic Rules and Thermodynamics. Of course, Thermodynamic Methods do not fail, but the Heuristic Rules are by nature only approximations that have a limited validity. One example in Process Integration is the so called (N-1) rule for the fewest number of units (heat exchangers). Since this formula only counts the number of process streams and utility types, with no reference to temperature, it sometimes fails to properly identify the correct minimum number of units. The calculation of minimum total heat transfer area is based on the assumption (or Heuristic Rule) that vertical heat transfer minimizes total heat transfer area. This assumption is not valid when there are significant differences in the film heat transfer coefficients for the streams. In the design phase, Pinch based methods fail to properly address the multiple trade-offs involved due to the sequential nature of these methods. The Pinch Design Method is also quite time-consuming, and even though the matching rules are simple, it often becomes a major effort to develop a valid initial design. The strict Pinch decomposition has also been shown to be counter-productive, since the subsequent design evolution is trapped into the structure of the initial decomposed design. Last but not least, another more important example is the plus/minus principle for the process modification that fails to account systematically and without enumeration for best combinations of process parametric and structural modifications for the sake of energy saving. In summary, there are limitations in many phases of Pinch Analysis, such as the problem definition phase (hard to handle forbidden matches), the targeting phase (approximations and heuristic rules that fail), as well as the design and optimization phase (multiple tradeoffs, topology traps, etc.). In theory, Mathematical Programming overcomes all these limitations however, some of the corresponding models are extremely difficult to solve. Finally, it should be mentioned that Mathematical Programming provides a framework for Automatic Design, which means that time (which is a limiting factor in many engineering projects) can be saved and used for more high level decisions.
Page 174 of 208
Classes of Mathematical Programming Models Generally, a Mathematical Programming model consists of an objective function (typically some economic criteria) and a set of equality constraints as well as inequality constraints. The general form is indicated below: min f(x,y) subject to g(x,y) 0 h(x,y) = 0 where x Rn y [0,1] m It should be noticed that the variables x and y in general are vectors of variables, and that the constraints g and h similarly are vectors of functions. The objective function (f) is assumed to be a scalar. If there are no binary variables (dim(y) = 0), and all functions f, g and h are linear, we have the simplest class of problems, the Linear Programming (LP) models. Using the Simplex algorithm, for example, LP models with hundreds of thousands variables and constraints can be solved in reasonable times with todays computer resources. If there are no binary variables (dim(y) = 0), and at least one of the functions f, g and h are nonlinear, we have a Non-Linear Programming (NLP) problem. These are generally much harder to solve, especially if the non-linearities are non-convex, because a local optimum may be found. If there are binary variables in the model (dim(y) > 0), and all functions f, g and h are linear, we have a Mixed Integer Linear Programming (MILP) problem. These can be solved to global optimality provided the number of binary variables does not cause a combinatorial explosion. Finally, if there are binary variables in the model (dim(y) > 0), and at least one of the functions f, g and h are non-linear, we have the hardest class of problems; Mixed Integer Non-Linear Programming (MINLP) models. Unfortunately, most real design problems are of the MINLP type with significant problems related to computer time (combinatorial explosion) and local optima (non-convex nature). Fortunately, in most Process Design, Synthesis and Integration applications, the binary variables do not occur in the equality constraints (h(x) = 0) and they appear linearly in the objective function and the inequality constraints.
Page 175 of 208
Rigorous Targets for Heat Integration As mentioned before one motivation for the use of Mathematical Programming in Process Integration is that rigorous targets can be obtained for energy requirement, number of units and heat transfer area, even in cases where there are restrictions on the matches. As indicated in the beginning of this chapter, the breakthrough in the use of Mathematical Programming in Process Integration came when results from Pinch Analysis were included in the models. The most important concept in this respect was the use of the heat cascade to formulate Transshipment Models. In these models, the Temperature Intervals act as Warehouses (or intermediate storages) between the Sources/Producers (the hot process streams and utilities) and the Sinks/Consumers (the cold process streams and utilities). The transshipment model will be introduced in little details along small part of the mathematical programming models for HEN in later chapters. Merge of Pinch Technology and Mathematical Programming It is the merge that has been done between Pinch and Mathematical Programming models, where thermodynamic Concepts and insights of pinch technique are combined with mathematical programming into what appears to be very powerful semi-automatic methods for both grassroots and retrofit heat exchanger network synthesis. This is the topic of the rest of this document. When using Mathematical Programming in energy Integration, there are three distinct activities that are of importance for the final result: 1) The structural alternatives must be represented in a Superstructure that is rich enough to contain all promising solutions, but not too wide to create prohibitive combinatorial problems. 2) The Superstructure must then be converted into a Mathematical Model that trades off complexity and rigor with speed in the numerical solution phase. This modeling activity is almost an art, and the same problem can be formulated quite differently. 3) The numerical solution phase where powerful Algorithms are needed to efficiently and reliably identify the global optimum for the proposed models.
Page 176 of 208
Even though process engineers primarily should deal with the first two of these activities, history has shown that major contributions have also been made in the last activity. The brief insights that can be summarized from pinch technology are as follows:
Given a minimum temperature approach, the exact amount for minimum utility consumption can be predicted prior to develop the network structure. Based on the pinch temperatures for minimum utility consumption, the synthesis of the network can be decomposed into sub-networks. The fewest number of units in each sub-network is often equal to the number of process and utility streams minus one. It is possible to develop a priori good estimates of the minimum total area of heat exchange in a network. It is possible via trial and error to select best process conditions that result in better energy targets It is possible for a given process to get a priori good selection of utilities and configuration of utility plant
While these insights narrow down the alternative designs for a HEN, process conditions modifications and utility plants configurations considerably (via reducing the search space), by themselves they do not provide an explicit procedure for deriving the configuration of a HEN, utility plants and/or define optimal process conditions. In this part we will present algorithmic optimization approach for the synthesis of HEN, utility plants and simultaneous optimization of process plant conditions and heat integration. The insights obtained from pinch technology will be used in the development of these mathematical programming algorithms. Hence we call this approach the Pinch technology and mathematical programming hybrid method.
Page 177 of 208
3.1
Mathematical Programming for HEN Synthesis The HEN synthesis problem features a number of key difficulties that are associated with its handling:
The potentially explosive combinatorial problem for identifying the best pairs of hot and cold streams so as to enhance energy recovery; Forbidden, required and restricted matches; The optimal selection of the HEN structure; Fixed and variable target temperatures; Temperature dependent physical and transport properties; Different phases type streams; Different types of heat exchangers
There are two major synthesis strategies for the HEN; sequential optimization and simultaneous optimization. The former is the one that rely on Pinch technology and exploit the insights resulting from it in developing the mathematical formulation of the sequential optimization method. Therefore it is the one that is going to be introduced here in this document. This HEN synthesis method provides systematic procedures that allow the automation of this synthesis task. Three basic heuristic rules that are motivated by the insights of pinch technology will be used in the development of algorithmic methods based on sequential optimization. In particular, it will be assumed that an optimal or near optimal HEN exhibits the following characteristics respectively: 1. Minimum utility cost 2. Minimum number of units 3. Minimum investment/Area cost It is possible in general to have conflicts among these rules. Therefore, we will assume that rule 1 has precedence over rule 2 and rule 2 over rule 3. In this way, our objective will be to consider first candidate networks that exhibit minimum utility cost, among these the one that have the fewest number of units, and among these the one that has the minimum investment cost.
Page 178 of 208
HEN Sequential Synthesis Minimum Utility Cost using Transshipment Model
Representation of Heat Flows for Transshipment Model
Hot Streams
Cold Streams
Interval # 1
Interval # 2
Interval # 3
Interval # 4
Page 179 of 208
The basic idea of the minimum utility cost calculation using the transshipment model representation above is to:
Introduce variables for all potential heat flows Write the overall energy balances around each interval Write mathematical model that minimize the utility cost subject to the energy balance constraints
Now for the heating and cooling utilities minimization problem, let us go back to our small problem that we solved algebraically using pinch technology. Then, we write our objective function, we formulate our model/constraints using energy balance; and we solve the optimization problem using commercial software. Objective function
min min Minimize (3* 10 6 Qheating + 5* 10 6 Qcooling )*3600*8760
Define loads of heating and cooling utilities in each temperature interval and the surplus from each interval via developing temperature interval diagram, tables of exchangeable loads and un-balanced thermal cascade diagram. Instead of writing the model for minimum utility cost using variables as we will do in the next part of this document for the minimum matches problem, Let us use numerical numbers to make easy to understand and illustrate how it is easy to model for optimization environment.
Page 180 of 208
Now let us have another look on the temperature interval diagram of our old example.
T minimum = 10 K
T* 555 515
Interval
Hot Streams 560 H1 520
Cold Streams 550 510
2 385 375 310 305 295 3 4 5 6 H2 390 380 330 310 300 380 370 320 300 290 C1 C2
Hot Streams:H1; F1Cp1= 10 kW/K H2; F2Cp2= 5 kW/K
Cold Streams:C1; F1Cp1= 10 kW/K C2; F2Cp2= 5 kW/K
Page 181 of 208
The model formulation is then as we said above simply a heat balance around each temperature interval as shown numerically in the graph below:
Thermal Cascade Diagram (Un-Balanced)
min Qheating
0.0 1300 100 750 100 50
r1 r2
1 2 3
760 2470 210 1050 380 0.0
r3
4
r4
5
r5
6
min Qcooling
min Qheating + 0.0-760=r1
r1+1300-2470=r2 r2+100-210=r3 r3+750-1050=r4 r4+100-380=r5

min r5+50-0.0= Qcooling
Page 182 of 208
Now using LINGO software: Model:

min min Min = (3* 10 6 Qheating + 5* 10 6 Qcooling )*3600*8760; min r1- Qheating = -760;
r2-r1= -1170; r3-r2= -110; r4-r3= -300; r5-r4= -280;

min Qcooling -r5= 50; min Qheating 0.0;
min Qcooling 0.0;
r10.0; r20.0; r30.0; r40.0; r50.0; End
Page 183 of 208
The solution of the optimization problem using LINGO software (student copy available upon request from ESU/CSD) gives the following results: Objective value = $ 255757
min Qheating = 2620 kW min Qcooling = 50 kW
r1= 1860 kW, r2= 690 kW, r3= 580 kW, r4= 280 kW, r5 =0.0 kW These targets obtained using mathematical programming with pinch based insights give an idea about the potential of utility saving in the facility. Minimum utility cost with constraint Matches In practice it might not always desirable or possible to exchange heat between any given pair of hot and cold streams. This could be due to the fact that the streams are too far apart or because of other potential considerations such as control, safety, environment regulations or start-up. Hence, it would be clearly desirable to extend the above mentioned formulation to the case when we impose certain constraints on the matches. The most common would be to forbid the heat exchange between certain pairs of streams (to force the use of utilities on some of the streams). We could also think of requiring that a minimum or maximum amount of heat be exchanged between certain pairs of streams (e.g. forcing the use of utilities on some of the streams). The LP transshipment model mentioned above implicitly assumes that any given pair of hot and cold streams can exchange heat since there was no information as to which pairs of streams actually exchange heat. In order to develop an LP model formulation where we do have that information, we can consider the following two alternative models: 1. Transportation model where we consider directly all the feasible links for heat exchange between each pair of hot and cold streams over their corresponding temperature intervals. 2. Expanded transshipment model where we consider within each temperature interval a link for the heat exchange between a given pair of hot and cold streams, where the cold stream is present at that interval and the hot stream is either also present or else it is present in a higher temperature interval. The second method is preferred since it renders a small size LP problem.
Page 184 of 208
HEN Synthesis with Minimum number of matches In the previous part we introduced the minimum utility cost target and its formulation as an LP transshipment model. The solution of the LP transshipment model provides:
The required loads of hot and cold utilities, and The location of pinch point if any.
A useful target used to distinguish among the many HENs that satisfy the minimum utility cost is the minimum number of matches problem which can be stated as follows: Given the information obtained from solving the minimum utility cost target problem, loads of hot and cold utilities, location of pinch points and hence sub-networks, determine for each sub-network the minimum number of matches and the heat load of each match. In such problem definition we implicitly assume that the HENs that satisfy minimum number of matches are usually close to an optimal or near optimal total annualized cost solution. Basic Idea of MILP Transshipment Model The basic idea in the transshipment model for the minimum number of matches target is to model explicitly the potential heat exchange between all pairs of streams, excluding hot utilities to cold utilities with respect to;
Existence of each match, Amount of heat load of each match, and Amount of heat residual of each hot process stream/utility.
The potential existence of each match (ij) is modeled through the introduction of the binary variables yij:
y ij = { , If match (ij) takes place 1 {0, otherwise

Where i HP U HU , j CP U CU , and we exclude some matches (ij) i HU , j CU .
Page 185 of 208
The amount of heat load of each match (ij) is modeled through the introduction of continuous variables Qij and Qijk:
Qij = Qijk ,
k
Where Qijk is the heat exchanged in match (ij) at interval k and Qij is the heat load of match (ij) over all intervals of the sub-network under consideration. The amount of heat residual of hot process stream/utility existing at each temperature interval k is modeled via the continuous variables, Ri,k and Rk:
Rk = Ri, k
i
Where Ri,k is the heat residual of hot process stream/utility I out of temperature interval k, and Rk is the total residual heat exiting interval k. The pictorial representation of a temperature interval k is shown in graph below. We have a hot process stream H1, a hot utility stream S1 potentially exchanging heat with cold stream C1. In such case we have two potential matches (H1,C1) and (S1,C1)
Page 186 of 208
RS1,k-1
RH1,k-1
QS1,C1,k Q H1,k
H1 H
TI-k QH1,C1,k
C1
Q C1,k
RS1,k
RH1,k
The amount of hot utility S1 which enters at an upper TI is known from the minimum utility cost calculation problem but the heat load Qs1,C1,k is a variable. The residuals Rs1,k-1 and Rs1,k are variables too. The known fixed quantities are the Qhot,H1,k and Qcold,C1,k which represent the heat available from hot stream H1 and the heat needed by the cold stream C1, respectively, at interval k. The residuals RH1,k-1 and RH1,k are variables. The heat load QH1,C1,k is also variable.
Page 187 of 208
Having introduced the pictorial representation and the appropriate binary and continuous variables the basic building blocks of the model for the minimum number of matches target are:
Energy balances for each hot process stream and at each temperature interval k, Energy balance for each hot utility and at each temperature interval k, Energy balances for each cold process stream and at each temperature interval k, Energy balance for each cold utility and at each temperature interval k, Definitions of total residual flows at each interval, Definitions of heat loads of each match, Relationship between heat loads Qij and binary variables yij, Non-negativity constraints on continuous variables, Zero top and bottom residual constraints, Integrality conditions on yij ( yij=0 -1)
Page 188 of 208
The objective function involves the sum of binary variables representing all potential matches. Then, the mathematical model for the minimum number of matches target can be stated as follows: Min s.t.
H Ri, k Ri, k _ 1 + Qijk = Qik , j
yij
i j
k TI i HPk j CPk U CUk k TI j CPk
H Ri, k Ri, k _ 1 + Qijk = Qik , i HUk , j
Qijk = Q
i
C jk
, j CPk
i HPk U HUk k TI
i
Qijk = Q
i HPk k TI
C jk
, j CUk
Rk Rik = 0
i
i HPk U HUk k TI
Page 189 of 208
Qi, j =
kTI
Qijk
Lij y ij Qij U ij y ij Qijk 0, k TI ; Rik 0, k TI R 0 = Rk = 0

yij=0_1
i HP U HU , j CP U CU
It is important to note here after we explained how to write the model that the objective function is a linear sum of all yijs and simply minimize the number of potential matches. The energy balances and the definition constraints are linear constraints in the residuals and heat loads. The relationships between the continuous and binary variables are also linear since Lij and Uij are parameters corresponding to lower and upper bounds, respectively, on the heat exchange of each match (ij). It is important too to understand the key role of these constraints which is to make certain that if a match does not exist, then its heat load should be zero while if the match takes place, then its heat load should be between the provided bounds. We can understand that more from the if statement below: If yij=1 , then If yij=0
Lij QijUij
,then 0 Qij 0 and hence Qij=0
Page 190 of 208
In the above mentioned model formulation constraint we will distinguish the following cases: Case (1) No restriction on Matches In this case, the upper bound Uij of potential match (ij) is given by the minimum of the heat loads of streams i and j, that is,
H U ij = min Qik Q C jc kTI kTI

The lower bound Lij, is used so as to avoid small heat exchangers and in this case of no restriction we will have Lij=0. Case (2) Required Matches: In this case, if one match must take place then we set Yij =1, And hence eliminate one binary variable from the model. Case (3) Forbidden Matches: In this case if one match (ij) must not take place for instance, then this is equivalent to setting: Yij=0 And the upper and lower bounds should be Uij=Lij=0, Which will make the heat loads Qij zero.
Page 191 of 208
3.2
Simultaneous Process Optimization and Heat Integration
When we consider an existing or still on paper process we need to optimize it to be energy efficient. However energy is not the only item for the operating cost and its systems components are not the only items in capital costs. In fact from the operating costs point of view raw materials are most of the time the dominant cost. Besides there are other process objectives such as safety, operability, reliability, environmental compliance, etc., that need to be considered while optimizing our process to be energy efficient. Of course, it will be optimal to be able to consider all the process objectives in one model and optimize the process. Numerical difficulty does not allow this approach to happen at least for now. Therefore, process optimization and energy integration tasks are always done sequentially. However, in some special cases where the problems are not large and only changes in a fixed process configuration are considered process optimization and heat integration can be done simultaneously. This situation is the one introduced here in this best practice. The pinch-based method introduced in this document is known as Pinch location method. In general, we describe our processes with models having nonlinear performance equations where pressures and temperatures are also variables. The main difficulty that arises is that we can no longer apply the equations of the transshipment model directly since the temperature intervals used in the previous model formulation will now be variable. Therefore, it is desirable to have a method that can handle both variable flowrate and temperatures. Let us first consider the non-linear optimization problem without heat integration. Here we will denote by x all the variables in the process among which are included the heat capacity flowrate(s) and the inlet and outlet temperatures, Fi, Tini,Touti,i=1,2,nh, fj,tinj,toutj, j=1,2,nc, of hot and cold streams respectively. The loads of the hot and cold utilities are denoted by Qs, Qw.
Page 192 of 208
The optimization problem corresponds then to:
Min C = f(x) + csQs+ cwQw s.t h(x)=0 g(x) 0
Qs = fj (t out t in ) j j
j =1
nc
Qw = Fi(Ti in Ti out )
j =1
nh
Qs, Qw 0, Fi, Ti in , Ti out 0, i = 1....nh, fj , t in , t out 0, j j j = 1....nc, x Rn

In this formulation, the objective term f(x), the equations h(x) = 0, and the constraints g(x) are in general nonlinear.
Page 193 of 208
In order to replace the Qs and Qw equations by heat integration constraints, it is essential to remove the definition of temperature intervals used in pinch method calculation procedures since these temperature intervals will not be fixed any more due to the possibility of changing inlet and outlet temperatures according to optimization needs of the process. Pinch Location Method Let us now assume first that we are dealing with the old problem where we have constant flowrate(s) and temperatures in order to explain how the Pinch Location method works without the need to use defined temperature intervals. Then we will incorporate in the above mentioned model formulation the necessary constraints that need to be added to account for heat integration without using the defined temperature interval concept. For the problem given in below we can determine the minimum utility consumption using methods mentioned before. Stream H1 H2 C1 C2 Fcp 1 4 2 0.5 Ts 450 400 300 360 Tt 350 350 360 500
Qs= 35 KW and Qw=145 KW and the pinch temperature using Tmin = 20 K is 450-430 K The pinch location procedures is very simple, we plot first the T-Q curves at a value of Tmin greater than 20 K. Try to pinch each of the inlet streams one-by-one as shown in the graphs below and determine the corresponding cooling and heating requirements. Clearly from the graphs below the pinch which is defined by inlet temperature of stream H1 is correct one. All others in graph exhibit temperature crossings and hence lower utility consumption (even negative). Therefore what the figures here suggest is that the criteria for selecting the correct pinch that can be used to define the minimum heating and cooling utilities that is feasible is to select the pinch that exhibits largest heating and cooling among all the pinch candidates.
Page 194 of 208
(A)
T (C) Pinch Candidate H1
500 H1 400 H2 C2
C1 300
Q (kW)
0.0
2000
4000
6000
Page 195 of 208
(B)
T (C) Pinch Candidate H2
500 H1 H2 400 C1 300 C2
Q (kW)
0.0
2000
4000
6000
Page 196 of 208
(C)
T (C) Pinch Candidate C1
500 C2 400 H2
H1
300
C1
Q (kW)
0.0
2000
4000
6000
Page 197 of 208
(D)
T (C) Pinch Candidate C2
500 C2 400 H2 H1
C1 300
Q (kW)
0.0
2000
4000
6000
Page 198 of 208
Mathematically, this condition can be expressed as follows:
{ } Qw = max{Q }
Qs = max Qsp
pP pP p w
Where P is the index set of all the hot and cold streams, i=1nh and J=1..nc,
p Q sp , Q w
, are the heating and cooling loads that result from each pinch candidate.
We can simplify these two expressions via calculating Qw from Qs as follows: Qw= {(heat content of hot streams- heat content of cold streams)} + Qs i.e. Qw =heat surplus () + Qs In other words if we know Qs we can then get the Qw directly without the max relationship. Therefore, our basic criterion for the pinch location can be expressed as follows:
Qs = max Qsp
pP
{ }
Qw = + Qs
Q sp in terms of flows and The only remaining part now is the explicit representation of
temperatures.
P P Q sp = { QAC QAH }
Page 199 of 208
Where QAC pinch p P ,
P QAH
} represents the heat deficit that exists above the candidate
P P QAC and QAH are the total heat content above the candidate pinch (p) of the cold and
P P of the hot streams, respectively. To develop explicit expression of QAC and QAH we can use the graphs below that show how we can calculate the heat content above pinch of hot stream (i) for different cases.
T
Heat content Above the Pinch Temperature T-pinch =
Fi (Ti in Ti out )
Ti in Ti out
Tpinch
Enthalpy
Page 200 of 208
T
Heat content Above the Pinch Temperature T-pinch =
Fi (Ti in Tpinch)
Ti in
Tpinch
Ti out
Enthalpy
Page 201 of 208
T
Heat content Above the Pinch Temperature T-pinch = 0
T-pinch
Ti in Ti out
Enthalpy
We can clearly see from the graphs above that the heat content of this hot stream above the pinch depends on whether the stream is entirely above the pinch, whether it crosses the pinch or whether it is below the pinch. And the three equations that capture the three cases are as follows: 1. Stream lies above the pinch, Ti in > Ti out > T p Then,
Fi[{Ti in Ti p } {Ti out Ti p }] = Fi[Ti in Ti out ]

2. Stream crosses the pinch, Ti in > T p > Ti out Then,
Fi[{Ti in Ti p } 0] = Fi[Ti in Ti p ]
Page 202 of 208
3. Stream lies below the pinch, T p > Ti in > Ti out Then, Heat content above pinch temperature = 0 From the above three cases the heat content above pinch (p) for hot stream (i) can be represented with the following formulation: Heat content above pinch (p) for hot stream (i) =
Fi[max{0, Ti in Ti p } max{0, Ti out Ti p }]

And the Total heat content of all hot streams above the candidate pinch (p), will be
P QAH
QA = Fi[max{0, Ti in T p } max{0, Ti out T p }]

P H i =1
nH
Similarly,
P C
P QAC
will be given by
QA = fj[max{0, t out (T P Tmin )} max{0, t in (T p Tmin )}] j j

j =1
nC
Where the pinch temperature, T P are defined as follows:
T P = Ti in
if candidate pinch is hot stream i
t in + Tmin if candidate pinch is cold stream j j
Based on the equations developed above where we obtained explicit expressions of the heat integration in terms of continuous variables such as flowrate and temperatures we can now write the modified process optimization formulation problem in a way so as to perform simultaneous process optimization and heat integration.
Page 203 of 208
The formulation is as follows
min C = f ( x) + csQs + cwQw s.t. h( x ) = 0 g ( x) 0 Qs fj[max{0, t out (T p T min)} max{0, t in (T p T min)}] j j

j =1 nC
Fi[max{0, Ti in T P } max{0, Ti out T P }]

i =1
nH
pP Qw = Qs + Fi(Ti Ti ) fj (t out t in ) j j
in out i =1 j =1 nH nC
Qs, Qw 0, Fi, Ti in , Ti out 0, i = 1...nH , fi, tiin , tiout 0, i = 1...nC , x Rn where T p, p P

TP=
Ti in
in j
if candidate pinch is hot stream i
t + Tmin if candidate pinch is cold stream j
Page 204 of 208
The above new formulation can treat the flows and the temperatures as variables for the optimization and the heat integration of any process. The difficulty in this formulation is the presence of max operators that are non-differentiable. However a smooth approximation procedure can be used to help avoids the difficulties and enable us use the current available commercial software solvers for non-linear programming problems as we are showing below.
Plot of max{0,f(x)}
f(x)
The function (x)=max {0,f(x)}, which arise in the formulation above is non-differentiable at f(x)=0 as shown in graph above. We can however, construct an approximation to it that is continuous and differentiable everywhere to avoid problems in the commercial NLP solvers. One very famous approximation to the (x) = max {0,f(x)} function is the following suggested by Balakrishna and Biegler.
Page 205 of 208
The function is simply replaced by the following equation:
a = 0.5[ f ( x) 2 + 2 ]1 / 2 + 0.5 f ( x)
It is recommended to use typical values for It is easy to verify that for small values of similar to the graph below.
between 0.0001 and 0.01.
the above equation yields an approximation
Plot of smooth approximation scheme
f(x)
Another difficulty which is not addressed in this document is the inclusion of the process structural optimization with heat integration in a mixed integer non-linear programming (MINLP) formulation for an industrial size application. This important problem will be addressed in another document.
Page 206 of 208
4.
4.1
Appendices
Computer Software(s)
The major state-of-the-art software(s) in the market for utilities targeting, selection and simulation are SUPERTARGET and Pro-Steam of KBC and Star of UMIST (copies are available in Saudi Aramco for UMIST software) The list of commercial and academic software for heat integration applications include the following, ADVENT, is process integration software based on pinch technology. It includes targets, design and optimization capabilities for heat exchanger networks. It also includes modules for utility system. It also performs exergy analysis using graphical method MAGNETS, is HEN design software that implements sequential synthesis strategy using the LP and MILP transshipment models, as well as, NLP superstructure optimization. SPRINT, is software for simulation, optimization, control and flexibility of heat exchanger networks. HX-NET, is a HEN design software. It includes targeting capability and is currently incorporated with ASPEN-Pinch software. SUPERTARGET, is a pinch analysis based software for targeting, design and optimization of HENs. It can be used for grassroots and retrofit problems. It has exergy analysis capability. 4.2 References
The two papers below present a brief but complete description of the state-of-the-art method in utilities targeting known as Total-Site-Analysis (Site-Pinch)
Linnhoff, B. and Dhole, V. Total Site Targets for Fuel, Cogeneration, Emissions and Cooling Kokossis, A. and Mavromatis, S. Conceptual optimization of utility Networks for Operational Variations-I. Targets and level optimization
Page 207 of 208
There are also several papers about Pinch technology and its application. Below are few important ones.
Linnhoff, B. and Townsend,B., Designing Total Energy Systems Linnhoff, B. and Townsend,B., Overall Site Optimization By Pinch Technology Linnhoff, B., Pinch AnalysisA State-of-the-Art Overview Linnhoff, B. and Smith, R., Recent Developments in Pinch Analysis
Regarding books there are variety of books that address pinch technology both in energy and mass integration applications. Below find few examples.
Smith, R., Chemical Process Design and Integration Biegler, L.,Grossmann,I. and Westerberg, A., Systematic Methods of Chemical Process Design El-Halwagi, M., Pollution Prevention Through Process Integration
22 April 2006
Revision Summary New Saudi Aramco Best Practice.
Page 208 of 208

Sabp A 009

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Sabp A 009

Încărcat de

Drepturi de autor:

Formate disponibile

Best Practice

Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

173 178 192

Appendices 4.1 4.2 Computer Tools References

207 207 207

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

Net Heat Sink Above the Pinch

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

Ts(C) 270 220 50 160 220 130 250 200 150 15

Tt(C) 160 60 210 210 220 130 250 200 150 20

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

STEAM H1 720 kW 180 kW 720 kW 360 kW

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

Q HP, out WHP

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

Grand Composite Curve (GCC) Without the Distillation Column STEAM

T0= 270C Q h, min= 1000

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

Original hot streams are hotter than proposed hot segments

Original Cold streams are colder than proposed cold segments

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

150 100 200 T

Look for the temperature that is beneficial To the energy targets

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

What is the cost of separating A from B

Energy Consumption= 200 HEN Area= 120

Energy Consumption=160 HEN Area= 132

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

Net Heat Sink Above the Pinch

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

Temperature (T)- Enthalpy (H) Diagram

Hot composite curve

Cold composite curve

SABP-A-009 Pinch Technology for Energy Efficiency Optimization

Temperature (T)- Enthalpy (H) Diagram

Hot composite curve

Cold composite curve

Capacity of C1, kW Capacity of C2, kW 19(550-510)= 760 0.0(550-510)= 0.0

Tstack Tambient Stack loss Fuel Enthalpy Q-Hmin