Ammonia Synthesis and Refrigeration















Project 3:


Heat Exchangers




















Group 3

Project Leader: Brandon Fleet

Project Members: Crystal Fontenot

Paul Kim



The Fubar Design Team modeled an improved version of Voltronís purification section of a 1000 metric tons per day ammonia plant using Aspen and Matlab. The new design is optimized to conserve energy, eliminate dependence on outside utilities, and reduce operating costs through the effective use of heat exchangers. To accomplish these requirements, an air-cooled heat exchanger, two standard counter current heat exchangers, and an innovative ammonia refrigeration loop is used. The system was modeled in Aspen and found to save $12,864,663 over the Voltron model in combined fixed and operating costs.



On November 5, 1997, Voltron, Inc. presented a design for the refrigeration and purification portion of an ammonia plant that produces 1000 metric tons of ammonia per day. Although the Voltron design was realistic, innovative, and cost effective, it was not energy efficient. The root of this problem lies in the designís ineffective use of heat exchangers. "Voltron, Inc. does not specialize in the construction and design of heat exchangers, a simplified design is provided in which the series of heat exchangers is combined into a few ideal heaters and coolers." [1]

Fubar, Inc. took on the task of optimizing Voltronís ammonia refrigeration and separation system, making the proposed design more energy efficient. This is achieved by implementing a system of heat exchangers that minimizes the use of utilities. Use of an air cooled heat exchanger eliminates the need for cooling water. Instead of ideal heaters and coolers, the Fubar design uses countercurrent heat exchangers that take advantage of other process streams to achieve the desired temperature changes. A refrigeration loop is incorporated that uses the plantís main product, ammonia, to cool the reactor effluent for separation. The new design significantly lowers fixed and operating costs of the plant and has been hailed as an "innovative design [that] showed creativity and made [the] system realistic." [2]




Voltron Model

The project requirements call for the optimization of an existing design proposal ñ familiarization with the original Voltron design is essential. A process flow diagram Voltronís design is included as Fig. 1


Fig. 1

The Original Voltron Design

All heating and cooling requirements are easily identified by the presence of "heat exchangers" in Fig. 1. The effluent from the reactor (stream 3) must be cooled from 377 C to -23 C before it enters the first flash tank (stream 4). It is the cooling of this stream that poses the greatest challenge in this project. The vapor from the first flash must be brought down to -27.5 before it enters the second flash. Finally, the combined liquid streams from both flash tanks must be heated to -5 C before they enter the final flash tank. To help meet these heating and cooling requirements, several other process streams with less stringent temperature requirements were utilized in the Fubar design. Characteristics of each of the streams important to our design considerations can be found in tabular form in Table 1.


Table. 1

Characteristics of Important Streams (Voltron Model)

Stream (see Fig 1.)
Flow Rate (Kg/hr)

(MW/ C)

Starting Temp. ( C)
Desired Final Temp. ( C)




















MATLAB Modeling

Once streams 3, 9, 13, and 16 (figure1) were identified as possible heat transfer streams, MATLAB was used to determine the extent to which each stream could be used in the heat integration of the plant.

First, the product of the mass flowrate and heat capacity of each stream was calculated using a matlab program called fcps.m (see appendix). Next, the ceng403 matlab program enthc1.m was used to figure out how much heat duty the cold streams in the system could remove from the hot stream. The cooling capacity of one of the streams, stream 13, was limited by itís own heating requirement. The other cooling streams were limited only by their F*Cp values. It was found that 7 MW of the required 43 MW could be removed from the hot stream by exchange with the cold streams. The other 36 MW would have to be handled by utility streams. This information was determined based on the graph produced by enthcl.m and is displayed in Fig. 2.


Fig. 2

In addition, a heating/cooling curve for each cold stream was produced using the ceng403 matlab program enthc.m . From it, the heat duty transferred was found. To aid in the developing of our Aspen model, a trial area for each heat exchanger was calculated using the general heat transfer equation:

Q = UAD Tln



Knowing the basic heat requirements of the Voltron design made it possible to create a system of heat exchangers that optimized the system for energy conservation. The first step in our optimization was to model the reactor as adiabatic. This is more realistic than the model presented by Voltron which used a simple rstoic reactor. As will be explained, the modeling of the reactor as adiabatic allowed Fubar to eliminate Voltronís second flash tank from the design. Another major change was the addition of an air-cooled heat exchanger (ACHE) right after the reactor that cools the hot stream down to 350 K. Two counter current heat exchangers, the first modeled rigorously, cool the hot stream even further. A final heat exchanger immediately before the first flash tank cools the hot stream down to the required -23 C by use of an innovative ammonia refrigeration loop. The process flow diagram for the Fubar design is included as Fig. 3.


Fig. 3

The Fubar Process Flow Diagram


The Reactor

The decision to model the reactor as adiabatic proved to be exciting on several levels. Simply put, it is a better model of how our unique system would operate if it were ever actually built. In the Voltron model, "The reactor was modeled in Aspen as an rstoic reactor, using the conversion of ~19% commonly achieved in industry." [1] Fubarís adiabatic model calculates how the reactor would perform under our specific design conditions instead of simply using an approximate conversion based on industry. When the adiabatic model was implemented, the stream flow rates and compositions changed in such a way that the second flash tank was no longer necessary to yield a high purity product. This had the effect of removing streams 6, 7, and 11 from the Voltron model and connecting stream 5 directly to the splitter leading to purge and recycle streams 8 and 9 respectively. The effect the adiabatic reactor had on the product stream is summarized in Table 2.


Table 2

Effect of Adiabatic Reactor Model on Product Stream



with second flash tank


with second flash tank


without second flash tank

Flow Rate (KG/hr)




Product Purity

(Mole frac. NH3)




The second flash unit was doing nothing to increase the purity of the final product stream, so it was eliminated from the design. By changing our model to the more accurate adiabatic reactor, we saw a large increase in product yield with only a small decrease in product purity.


The Air-Cooled Heat Exchanger (ACHE)

A majority of the cooling of the process stream is done with an air-cooled heat exchanger placed directly after the reactor. An air-cooled heat exchanger was chosen because of its many advantages over water-cooled exchangers Water-cooled exchangers can lead to thermal and toxic pollution of natural water sources. Air-cooled heat exchangers have very little impact on the environment, preventing the problem of possible water pollution. Also, unlike water, air is available in unlimited supply and has no procurement or waste disposal cost. With air, there are no worries about salt, scale, calcerous or mud deposits, and there is no growth of biological substances in the system.

When deciding whether an air-cooled heat exchanger can be used in a plant, several factors must be taken into account. If the process stream needs to be cooled to below ambient air temperature, air cooling cannot be used alone. To be economically sound, the temperature difference between the outlet process stream and the ambient air needs to be in the range of 10-15C (18-27F). [3] For example, an air-cooled heat exchanger operating in a plant on the Texas gulf coast can only cool the process stream to about 112F. The cooling requirement of the process stream in the Voltron design dictated that the stream be cooled to 77 C (170.6F). This was within the capabilities of an air-cooled heat exchanger.

Once the feasibility of an air-cooled heat exchanger was established, care was taken to determine the optimum design of the exchanger. A tube bundle length of 30 ft was chosen to satisfy the specification that there should be at least 40% fan coverage. [3] The number of tube bundles was set at six., consistent with typical designs for gas coolers. [3] As with most air-cooled heat exchangers, the tubes are finned.


The Heat Exchangers

Although the air-cooled heat exchanger is responsible for most of the cooling in our process stream, an additional three counter current heat exchangers are used to cool the hot stream down to -23 C. They are strategically arranged in such a way that the heat from various process streams is exchanged to minimize energy waste.

The first of these heat exchangers, HX-1, cools the hot stream from 77 C to 47 C using recycle stream 9. As required by the rules of this project, HX-1 was modeled rigorously in Aspen. HX-1 was chosen to be rigorously designed due to its complexity and its central role in our design. This exchanger is much larger and far more complex than HX-2 and absolutely central to any network exchanger design. Of considerable concern in our design of this exchanger was the difference in pressure between the hot and cold streams.

This exchanger was deliberately placed prior to the compressor in order exploit the pressure difference between the streams. This would facilitate maximum heat flow with a minimum required area. The area was determined by modeling a shortcut exchanger first. Use of this area enabled the selection of a typical tubular heat exchanger geometry. [4] A two pass tube bundle geometry of 1044 16í æ" tubes with a triangular 1" pitch within a 37" shell of TEMA type F was selected. Rigorous modeling revealed excellent agreement with our shortcut modeling.

However, ASPEN encountered difficulties associated with the large operating pressures required by the system. It was necessary to develop MATLAB programs capable of determining the required pipe thickness to withstand the operating internal and external pressures. The tubes operated under the greatest pressure differential with internal pressures 1x107 and external pressures of 1x10 6. With safety factor of 2, a tube wall thickness of .1718cm was determined which coincided with .18cm tube wall thickness given by a BWG of 7.

Pressure drops over the heat exchanger lengths were also considered in the rigorous modeling of HX-1. Tubes of 1" diameters were also modeled but it was felt the resulting increase in shell size offset potential advantages. Furthermore, with considerable tube side pressures, it was felt that the flow could sustain the pressure drop associated with its passage through the exchanger. Higher costs were associated with use of a high pressure heat exchanger and cost correlation were used to determine HX-1 purchase costs. [5]

The second heat exchanger, HX-2, exists primarily to heat up process stream 12 so it may enter the final flash tank at the proper temperature. The cold stream is heated from -18 C to -5 C by the hot process stream. Due to the relatively small flow rate of stream 12, HX-2 is much smaller than HX-1; with a surface area of 24m2 as opposed to 305m2. The cooling effect on the hot stream in HX-2 is minimal.

The third and final heat exchanger, HX-3, brings the process stream down to -23 C by exchanging it with 92%-liquid ammonia at -33 C. The unique characteristics of the ammonia refrigeration loop provide for enough heat transfer to lower the hot stream the final 67 degrees down to -23 C.


Ammonia Refrigeration Loop

Despite the efforts of the air cooled heat exchanger and the chilly recycle stream in HX-1, the reactor product stream still remains at 44oC with only one cold stream left for exchanging. This cold stream, the final product stream, was found to have a mass flow insufficient to reduce the hot stream to -23o C. In order to increase the mass flow of the final product stream without disturbing the conditions of the existing streams an additional recycle loop was introduced.

The final product stream was used to cool the hot stream after which it was flashed and the 35% bottoms (liquid) stream was reintroduced to the final product stream (liquid) after Flash 2. However, the introduction of the post HX-3 flow into that of the final product stream warmed the cold stream of HX-3. In order to maintain that stream at ~238o K and significantly reduce its enthalpy it was necessary to throttle the combined recycle/product liquid stream into an 8% vapor-liquid composition. Prior to throttling it was necessary to pressure the stream to 9x106 N/m2.

This proved to be an economical feasibility due to the prior elimination of a flash vessel in the earlier design stages and the reduced utility costs of cooling the hot stream. Furthermore, the additional flash vessel is significantly smaller than the removed vessel, processing a less than half the mass flow of the removed vessel. The most significant utility cost is incurred in pressurizing the stream prior to throttling, but use of a liquid stream insures that the pump costs are relatively low. Unfortunately the use of a mixed composition cold stream significantly increases the required surface of HX-3. However, the surface area is not unreasonable for practical placement of a bank of exchangers and is justified by the overall decrease in operating costs.



The ultimate justification for implementing any new system is a net savings in cost over time. The Fubar model is no exception; every aspect of our proposal was designed with the dollar in mind. By examining the savings in both fixed and operating costs, it is possible to get an idea of the magnitude of savings that can be achieved with carefully designed heat exchangers.


Fixed Costs

The reactor recycle compressor compromises bulk of the fixed cost in both the Voltron and the Fubar design proposals. However, the Fubar design has a smaller flow rate in the reactor recycle stream, making installed costs of the compressor slightly less. The installed costs of the Fubar heat exchangers are generally less than the giant heaters and coolers used in the Voltron design, however the cost of the air cooled heat exchanger balances out the total installed cost for all heat exchangers. Fubar eliminates the second flash drum from the Voltron design, replacing it with a smaller flash for the ammonia refrigeration loop. Also included in the Fubar proposal is the added cost of an additional pump in the refrigeration loop. As can be seen in Table 3, the fixed costs of the Fubar design are lower than the Voltron model, although not by a significant amount.


Table 3


Estimated Fixed Costs



















Reactor Air Cooler








Recycle Cooler








Flash Bottoms Cooler








Product Cooler








Reactor Recycle Compressor








Product Recycle Pump








Flash Drum 1








Flash Drum 2








Flash Drum 3






Total Costs






Operating Costs

Operating costs are far more important than fixed (installed) costs of a plant. Savings from a fixed cost can determine whether a company has the capital available to start up a plant, but savings in operating costs continue to benefit the company year after year. Operating costs were calculated assuming plant operation 350 days out of the year and an electricity cost of $0.06 per Kw-hr. The first point of interest is that the new design actually produces more product than before. Although his should amount to a higher profit margin, it was not factored into the operating cost savings. The majority of savings comes from the elimination of utilities in the Fubar design. By not having to purchase steam, refrigerant, or coolant water, the operating cost of the heat exchangers is minimal. The only cost is from the electricity it takes to drive the fans, which was conservatively estimated at 1% of their duty. A detailed operating cost comparison appears as Table 4.


Table 4


Estimated Operating Costs













mass %






Compression Costs


Voltron (KW)






















Heat Exchanger Costs*








usage (kg/hr)






price ($/kg)








usage (kg/hr)






price ($/kg)








usage (KW)












Operating Costs:







Fubar, Inc. has presented a well thought out, innovative, and cost effective design that improves on the Voltron system by making it more energy efficient. There are several key advantages that make the Fubar proposal attractive. The reactor is modeled as adiabatic, bringing a higher degree of realism to the Aspen calculations. The Fubar design uses no outside utilities. The environmentally friendly air-cooled heat exchanger eliminates the need for coolant water. Using a recycle in the product stream allows Fubar to benefit from an increased flow rate of the final cooling stream without having to purchase outside refrigerant. The other heat exchangers take advantage of process streams throughout the plant to perform heat exchanges. Furthermore, Fubar has a lower installed cost, and more importantly a much lower operating cost. The engineers at Fubar are very excited about their proposal and hope that others will share in their excitement after reviewing these facts about the design.



[1] Liu, V., Project 2 Written Report: Voltronís Refrigeration System of an Ammonia Plant, November 1997, p. 3, 5

[2] Undisclosed Authors, Class Comments on Group 3 Oral Presentation, Dec. 9, 1997

[3] Mukhergee,R., Effectively Design Air Cooled Heat Exchangers, Chemical Engineering Progress, February 1997, p. 27, 37, 40

[4] Peters and Timmerhaus, Plant design and economics for chemical engineers. McGraw-Hill, 1991

[5] Fixed Capital Cost ñ A Preliminary Estimate

[6] Tubular Exchanger Manufacturers Association, Standards of the Tubular Exchanger Manufacturers Association Seventh Edition,1988

[7] Rice University CENG 403 homepage,




function Fcps = fcps(T,state,stream)

%function Fcps = fcps(T,state,stream)

%Finds Fcp for a given T for all streams in a system.

%T= Temperature in Tdeg

%state = state of stream: 'l' for liquid, 'v' for vapor

global cpv cpl ns ne Fcps

ourns = ne(stream, 4:length(ne));


if state == 'v'

tvec = T.^(0:4);

cpvec = cpv*tvec';


tvec = T.^(0:3);

cpvec = cpl*tvec';


Fcps = sum(ourns'.*cpvec*1E-3);