Pinch analysis used in retrofit design of distillation units

June 21, 1999
A substantial reduction in energy consumption can be achieved at the Petroleos Mexicanos (Pemex) refinery of Tula, Hidalgo, Mexico, by retrofitting one of two distillation plants. As part of a larger program to reduce energy consumption in Mexican refineries, Pemex commissioned Instituto Mexicano del Petroleo (IMP) to identify opportunities for reduced energy consumption in the refinery of Tula, Hidalgo.
Victor Briones
PMI Holdings North America Inc.

Ana L. Pérez, Luz M. Chávez, Rubén Mancilla, Marisol Garfias, Rodolfo Del Rosal, Nancy Ramírez
Instituto Mexicano del Petroleo
Mexico City

A substantial reduction in energy consumption can be achieved at the Petroleos Mexicanos (Pemex) refinery of Tula, Hidalgo, Mexico, by retrofitting one of two distillation plants.

As part of a larger program to reduce energy consumption in Mexican refineries, Pemex commissioned Instituto Mexicano del Petroleo (IMP) to identify opportunities for reduced energy consumption in the refinery of Tula, Hidalgo.

As a result of this study, IMP proposed a retrofit of the crude-distillation unit to reduce fuel consumption by more than 40%. This will save about $8 million/year and result in a payback period of less than 2 years.

Implementation of IMP's suggestions is on hold, pending development of the economic environment of the refining industry.

Using pinch-analysis techniques for energy analysis, IMP's design considers heat integration between the atmospheric and vacuum distillation units. The original design did not consider this heat integration.

The objectives of the redesign were to maximize distillate yields, meet stricter product specifications, and reduce energy consumption.

IMP used a method proposed by Liebmann1 to simultaneously analyze the atmospheric and vacuum distillation units and the heat-exchanger network. The authors also addressed trade-offs between heat recovery in the network and column hydraulics when defining the pump-around temperature.

Unit configurations

The crude-distillation unit at the Tula, Hidalgo, refinery consists of the atmospheric distillation unit, which processes crude oil, and the vacuum unit, which processes atmospheric residue.

The atmospheric distillation unit was designed to process 170,000 b/d of a blend of Maya crude oil (22° API) and Isthmus crude oil (33° API) under two operating modes: maximum gasoline yield or maximum diesel production.

A simplified process flow of this unit is shown in Fig. 1 [92,302 bytes].

The crude is preheated with product streams and atmospheric residue in a heat-exchanger network (Fig. 2 [78,683 bytes]).

In an intermediate part of the heat-exchanger network, crude desalting occurs at 140° C. After desalting, the crude temperature is increased to 214° C. before it enters two topping columns where the light naphtha and light ends are removed. The overhead products are treated in a stabilizer plant.

The reduced crude from each topping tower is heated in a fired heater to 368° C. and fed to the atmospheric column. In the atmospheric column, the overhead vapor is condensed and divided into topping tower reflux, crude distillation reflux, and naphtha product.

Jet fuel, kerosine, light atmospheric gas oil (LAGO), and heavy atmospheric gas oil (HAGO) are side withdraws from the tower. The former three streams are stripped to eliminate light hydrocarbons and meet specifications.

In the existing design, stripping can be carried out using superheated low pressure (LP) steam or reboiling, which is provided by the hot atmospheric residue from the tower bottoms.

All products exchange heat with crude oil before receiving further treatment in other plants or being sent to storage (Fig. 2).

The original design of the atmospheric distillation column had two side-heat withdraws to be used in the crude oil preheat exchanger network, providing 40.5million kcal/hr. The heat withdraws are two pump-backs.

The vacuum unit processes 90,000 b/d of atmospheric residue. A simplified process flow is shown in Fig. 3 [84,517 bytes].

The charge to this plant is preheated by vacuum gas oil (VGO) and residue from the vacuum tower bottoms in a heat-exchanger network, then heated in two fired heaters to about 380° C. The products are light vacuum gas oil (LVGO), heavy vacuum gas oil (HVGO), and vacuum residue.

The vacuum column design has 10 theoretical stages. Three packing beds and four trays are in the stripping section of the tower. An operating pressure of 12 mm Hg obtains a 540° C.+ vacuum residue.

Design methods

In the past decade, several methods to design distillation columns and to reduce the energy consumption by pinch analysis have been reported. 1 2

In 1991, Dhole proposed a method that used principles of distillation, rigorous column simulations, and pinch-analysis techniques. Based on a column simulation, a grand composite curve for the distillation column is constructed.

The grand composite curve provides information about the best thermodynamic condition of the feed and optimum reflux use. The location and heat load of side reboilers, side condensers, and pump-arounds are also provided.

Although Dhole's method is useful for designing distillation columns, it requires a rigorous simulation of the complex column, which greatly depends on design criteria and the designer's experience.

In 1996, Liebmann reported a new design method, also based on principles of distillation, rigorous column simulations, and pinch-analysis techniques. He decomposed the complex distillation column into a sequence of simple distillation columns. The decomposition of columns allows easy simulation and convergence.

The method proposed by Liebmann explores trade-offs between reflux and the number of trays in all the separation sections of the column. It looks for the optimum number, location, and heat loads of pump-arounds. It also explores the trade-offs between the use of stripping steam and reboilers in the strippers.

The last step of the method analyzes the effect of thermal coupling between simple columns and the trade-offs between heat recovery and separation efficiency.

Although Liebmann presents a complete discussion about the design and retrofit of crude oil distillation columns, he does not address the definition of several design variables for crude-distillation units, which will be discussed here.

He did not discuss the definition of the minimum approach temperature (DTmin)3 for the design, the definition of the outlet temperature in a pump-around (i.e., the return temperature to the column), and the approach for designing heat integrated units.

Energy vs. investment costs

Liebmann used a DTmin value of 10° C. in all his examples. The design of a crude-distillation unit using such a small DTmin value can lead to a design with low energy costs but high investment costs.

It is important to analyze the trade-offs between energy cost and capital cost. Because this method involves the simultaneous analysis of the distillation column, the heat exchanger network and the fired heaters, the conventional targeting methods of pinch analysis for heat exchanger networks are not applicable.

One option could be to use the overall approach for three different DTmin values. Different approach temperatures for certain streams may also be used. First, a polynomial estimate for the minimum cost can be reiterated until the estimated cost at optimum is close to the actual cost. This option, however, can take a long time.

Some operation and process considerations can help define the DTmin in the design:

  1. If the crude-distillation units process a heavy crude oil, heat transfer coefficients are small as a result of the high viscosity of crude and vacuum residue streams. Thus, the heat transfer area can be too large if DTmin is small.
  2. If the crude-distillation units process a blend of crudes, it is possible to have large variations in the feed and operating conditions. Use of a small DTmin in the heat-recovery network can lead to operating problems in the plant.
  3. The use of stream-specific approach temperatures can be reliable if it is combined with the experience from heat-exchanger design experts.
The retrofit design of the crude distillation unit at the Tula Hidalgo refinery, was developed with a DTmin of 40° C. This DTmin value was defined in conjunction with heat-exchanger design experts and the above considerations.

Pump-around return temperature

Liebmann did not discuss how to define the pump-around return temperature and its effect on the design of the crude-distillation unit.

The examples reported in his research work are solved considering a return temperature around 60° C. less than the draw temperature. There are, however, important trade-offs that should be explored.

A high return temperature represents better opportunities of heat recovery and facilitates the design of the heat exchangers (as a result of a greater driving force). A high return temperature, however, can also increase the flow rate in the pump-around (for a given heat duty) and can cause flooding problems in the column.

Generally, the trouble is, when the whole analysis is started, that the duty for the pump-around in question is not known. As the analysis is initiated with no thermal coupling, the duties of all column pairs operating on one side of the pinch (that is, of the debottleneck)3 will be larger than their final values, and the duties of the column pairs straddling the pinch will be smaller than the final values.

One approach is to set the pump-around return temperature to the maximum that can be tolerated by hydraulics. Thus, the designer is able to see the consequences for the hydraulics even in the simple columns.

Then, the designer should fix the maximum pump-around flow and vary the return temperature when, during the design procedure, the duty decreases. This design procedure is complicated because by varying the pump-around duty, the designer will vary its return temperature.

The following steps can be used to define the pump-around return temperature to obtain the most heat recovery and avoid flooding problems in the distillation column:

  1. Begin the design with a high return temperature, and obtain the pump-around flow rate.
  2. Check heat recovery using the grand composite curve of the process.
  3. Check hydraulic analysis of the distillation column section. If there is any flooding problem for a given column, decrease the return temperature to reduce the flow rate and overcome the flooding problems.
Having an initial hydraulic analysis of the column, the maximum allowable flow rate can be specified and the pump-around return temperature adjusted. Then heat recovery should be checked again.

For a retrofit design, the above approach can be followed if an initial hydraulic analysis is not available. Otherwise, the maximum allowable flow rate in the distillation column section can be specified from the beginning, and then the pump-around return temperature can be estimated based on this maximum flow rate. Once both variables are specified, heat recovery and hydraulic analysis can proceed.

Integration of distillation units

Integration of the atmospheric and vacuum distillation units is logical when considering energy savings. The integrated design has several advantages:
  1. It allows additional opportunities for heat recovery.
  2. It removes inefficiencies such as the cooling then heating the same streams.
The integrated design uses a common heat exchanger network for the atmospheric and vacuum distillation units, so the approach proposed by Liebmann can be applied with the following steps:
  1. Simulate the simple columns sequence for the atmospheric column.
  2. Simulate the simple columns sequence for the vacuum distillation column.
  3. Build the grand composite curve considering all the process streams in both the atmospheric and vacuum distillation units.
  4. Explore the trade-offs between reboiling and steam-stripping in the atmospheric unit as no strippers are commonly used in the vacuum unit.
  5. Analyze the advantages of thermal coupling (benefit in separation efficiency vs. penalty in heat recovery) simultaneously for both, the atmospheric and vacuum distillation units.

Proposed modifications

The proposed revamp of Pemex's Tula, Hidalgo, refinery makes a deeper cut in the crude oil to increase gas oil yield, which can be processed in the fluid catalytic cracking unit (FCCU).

Fig. 4 [85,219 bytes]shows the proposed modifications to the atmospheric distillation section.

The proposal changes the tower internals. The atmospheric column will use structured packing. Some beds have high separation efficiency, while others have low separation efficiency but with high liquid capacity to provide the liquid flow of the pump-arounds.

To obtain a 565° C.+ residue, the vacuum distillation column will also be repacked, and its internal diameter will be modified in some sections.

The authors used several design criteria to design the revamp:

  • An overflash value of 2.5% is used for the atmospheric distillation.
  • A washing flow of 0.15 gal/min-sq ft is used in the low section of the vacuum tower to reduce metal and pitch entrainment in the VGO.
  • A temperature of 360° C. is maintained in the bottom of the vacuum tower recycling a cold residue stream to minimize residue cracking.
  • Medium pressure steam is used in the fired heaters to reduce cracking and coke deposition.

Strippers design

To produce multiple products from a single distillation column, the tower should have side draws and strippers. The objective of the strippers is to remove light components of the side draws to meet the product specification.

Energy must be supplied to the system by using either stripping steam, which is superheated low pressure (LP) steam, or by using a reboiler with a process hot stream as the heating medium.

Stripping steam requires heat input at lower temperatures (i.e., LP steam). It is suggested because it provides better heat recovery than a reboiler.

The use of a reboiler increases the stripper bottom temperature (a cold stream), but it also increases the product temperature (a hot stream). Increasing the temperature of a cold stream has a negative effect on heat recovery (keep cold streams cold), but increasing the temperature of a hot stream has a positive effect on heat recovery (keep hot streams hot).

Therefore, the use of a reboiler must be analyzed carefully so that the global consumption of heating and cooling utilities does not increase. Reboiling also reduces stripping steam consumption.

In this work, there was a small opportunity to use a reboiler rather than stripping steam in the jet fuel stripper. This option was dismissed, however, because a reboiler increased the heating and cooling utility consumption.


One of the main concerns in the analysis of the distillation columns is how to define the optimum number of pump-arounds, their heat load, and location.

According to the method reported by Liebmann, the decomposition of the complex column into a sequence of simple columns establishes an initial configuration that comprises the initial definition and location of pump-arounds. The number of pump-arounds is equal to the number of side draws.

For the case study, there were four side products (jet fuel, kerosine, diesel, and HAGO) and thus, four pump-arounds in the atmospheric distillation column.

In the vacuum tower, there were two side draws (for LVGO and HVGO), and thus, two pump-arounds.

An additional pump-around was located in the bottom of the vacuum column to quench the bottom product recycling cold vacuum residue. The quench maintains a bottom temperature in the tower of around 360? C. to reduce residue cracking.

The analysis of thermal coupling between simple columns leads to the final heat loads and temperatures of the pump-arounds (Table 1 [44,515 bytes]).

The installation of all these pump-arounds significantly increases the temperature and load of the heat sources, which provides better opportunities for heat recovery.

Recovering heat from the condenser

Recovering heat from a condenser increases system heat recovery. Two options were considered in the analysis: a one-stage condenser or a two-stage condenser.

A one-stage condenser is simpler, but the temperature of the heat source is lower.

The choice of condenser depends on the location of the hot stream (top vapors) in the process grand composite curve.

For example, if a condenser with one stage is used and the hot stream crosses the pinch, the use of a condenser with two stages should be considered since it increases the temperature of the hot stream (that is, heat is displaced from below to above the pinch). The result, then would be a reduction in heating and cooling consumption (according to the + or - principle3).

In this work, a condenser with one stage was proposed because the hot stream is located far from the process pinch, and a two-stage condenser did not reduce heating and cooling consumption.

There was a significant reduction in energy consumption, however, as a result of the integration of the condenser with the crude (a cold process stream).

Air preheaters

The grand composite curve of the process provides information about the optimum location of utilities.

Fuel gas with a dew point of 190° C. is used as the main heat source in the crude oil distillation unit. The process pinch is located, however, at a temperature of about 300° C., limiting the cooling of the hot flue gas. Thus, the stack temperature would be greater than 300° C., and the fired heater efficiency would be around 80%.

The installation of an air preheater in the fired heaters will cool the flue gas to 200° C. with a significant increase in furnace efficiency and savings in fuel consumption.

Benefits of the retrofit design

Increased product yields within new product specifications and energy savings are the main benefits in the design.

The design increases gas oil yield by 4,200 b/d. This additional gas oil production is hydrotreated and processed in a FCCU, which provides gasoline and olefins.

Increased olefins leads to increased production of MTBE, TAME, and alkylate. All these streams ultimately increase the gasoline yield of the refinery.

Several modifications were proposed to reduce the energy consumption. A summary of these modifications is:

  1. An integrated design of the atmospheric and vacuum distillation units exploited opportunities for heat recovery and removed inefficiencies. This consideration results in a reduction of the fired-heat load in the atmospheric and vacuum sections and a reduction of total fuel consumption.
  2. Redesign of the atmospheric distillation column with a larger number of pump-arounds with higher temperatures and heat loads improved heat recovery in the preheat train.
  3. Use of stripping steam instead of reboilers in the strippers of the atmospheric distillation column was beneficial. The furnaces use less fuel as a result of the use of heat sources (for example, vacuum residue and pump-arounds) exchanging heat with the crude oil (instead of providing heat to the reboilers).

    Fuel savings also result from an increase in the cogeneration opportunities in the steam and power plant. The stripping steam is obtained from the LP steam header of the refinery and superheated in the convection section of the fired heaters. This opportunity was identified by using Total Site techniques.4

  4. Heat recovery occurred in the condenser.

  5. The energy efficiency of the fired heaters improved. Additional fuel savings were obtained by air preheating systems. The stack temperature of the flue gas was reduced, minimizing heat losses and increasing energy efficiency.

Finally, the retrofit design of this crude unit offers additional profit as a result of the improvement in product yields and product specifications.


  1. Liebmann, K., "Integrated crude oil distillation design," PhD Thesis, University of Manchester Institute of Science & Technology, 1996.
  2. Dhole, V.R., "Distillation column integration and overall design of subambient plants," PhD Thesis, University of Manchester Institute of Science & Technology, April 1991.
  3. Smith, R., Chemical Process Design, McGraw Hill, 1995.
  4. Raissi, K., "Total site integration," PhD Thesis, University of Manchester Institute of Science & Technology, 1994.

The Authors

Victor Briones is technical manager at PMI Holdings North America Inc., Houston.

He worked for more than 15 years at IMP as a senior researcher, in which role he participated in the development of computer tools for process synthesis and optimization. Briones also headed a team which performed projects to reduce energy consumption in Pemex refineries.

Briones holds a BS from Instituto Polit?cnico Nacional and a masters degree in process engineering from Universidad Nacional Aut?noma de M?xico, both in Mexico City. He also holds a PhD in process integration from UMIST, Manchester U.K.

Ana L. Pérez heads the basic engineering development group at IMP.

She is involved in the development of tools and application of new methods for improving refining and petrochemical processes. She is also involved in the analysis, optimization, and planning of refining projects.

Pérez holds a BS in chemical engineering from Instituto Polit?cnico Nacional, Mexico City.

Luz M. Chévez is a specialized researcher at IMP. She works in the process synthesis and optimization area, in which she has had 11 years of experience.

Chévez applies optimization techniques on utility systems and on the synthesis of heat exchanger networks, distillation columns, and gasoline reformulation.

She holds a masters degree in chemical engineering from Instituto Polit?cnico Nacional, Mexico City.

Rubín Mancilla is head of the process synthesis and optimization group at IMP. He is responsible for projects related to energy optimization in the Mexican petroleum industry. He has more than 15 years of experience in the development of process synthesis and optimization tools at IMP.

Mancilla holds a BS in chemical engineering from Universidad Nacional Autónoma de México, Mexico City.

Marisol Garfias works at IMP in the process simulation group. She has participated in the simulation and optimization of refining processes as well as development of new routines for the process simulator at IMP.

Garfias holds a BS in chemical engineering from Universidad Nacional Autónoma de México, Mexico City.

Rodolfo Del Rosal is a process manager, responsible for the execution of basic engineering designs at IMP. He has more than 25 years of experience in research, development, and design of refining and petrochemical technologies and in the development of computer programs for process synthesis and optimization.

Del Rosal holds a BS in chemical engineering from Universidad Nacional Autónoma de México, Mexico City.

Nancy Ramírez is a process engineer at IMP in the synthesis and optimization group. She has participated in projects related to the optimization of utility systems and energy integration for distillation columns. She has also worked in retrofitting exchanger networks.

Previously, she was a lecturer at Universidad Nacional Autónoma de México, Mexico City. Ram!rez holds a BS in chemical engineering from Universidad Nacional Autónoma de México.

Copyright 1999 Oil & Gas Journal. All Rights Reserved.