Design of a heat exchanger

K. SHERWIN and M. MAVROMIHALES, School of Engineering, University of Huddersfield, Queensgate, Huddersfield, HD1 3DH, UK
Received 18th September 1997 Revised 10th February 1998 NOMENCLATURE A E h Nu Pr Q Re T Tm U surface area, m2 effectiveness heat transfer coefficient, W m2 K Nusselt number Prandtl number heat transfer rate, W Reynolds number fluid temperature, C mean temperature difference, K overall heat transfer coefficient, W m2 K

Subscripts a w 1 2 air-side water-side inlet conditions outlet conditions

INTRODUCTION The paper describes the designbuildtest of a heat exchanger as a final-year design project for undergraduate students in Mechanical Engineering. Design can be defined as the use of imagination, knowledge, experience and judgement to define an end product [1]. Much of this activity cannot be taught in the normally accepted sense like an engineering science. Students can only learn to design and gain experience by being actively involved in design. Hence the need to teach design through projects. It is important that students realize the goal at which they are aiming; the definition of an end-product. Many undergraduate design projects end up as paper studies. This is inevitable, owing to the constraints on time and finance within the curriculum. Unfortunately, paper studies do not provide students with the feedback on how well their designs work, or whether they will work at all. Therefore, at Huddersfield, a conscious effort is made to include some projects of a designbuildtest nature so that students do have direct feedback on the success, or otherwise, of their designs. One such project, used in the first year of the course, was the design of a compressed air engine [2].
International Journal of Mechanical Engineering Education Vol 27 No 3

210

K. Sherwin and M. Mavromihales

The use of a heat exchanger forms an excellent basis for a project. A design project involving the building and testing of a shell and tube heat exchanger has previously been described [3], whilst at Lancaster the building and testing of an air cooled heat exchanger has been used for many years to underpin a second-year course in heat transfer [4]. However, both these projects result in a limited range of geometries, being constrained, in the first project by the thermal resistance across the tube wall and in the second by a fixed tube length and diameter. Design is open-ended and the present project reflects this by allowing a wide range of geometries. The type of heat exchanger chosen is that in which hot water flows through a staggered bank of tubes and is cooled by a cross-flow of air. Even when the tube diameter is fixed the students must still determine: exchanger height exchanger width tube pitch number of rows

This provides a challenging design analysis which is combined with a simple form of construction that can be built in a short space of time. It also allows the integration of CAM into the design project. When the project was originally started, the final-year design course was based on two major group projects and students were allowed two design weeks to concentrate solely on each design project without any other formal teaching commitments. The design weeks provided the ideal opportunity for students to concentrate on the construction of their particular heat exchanger. Over the years, constraints on time and teaching meant that design weeks were reduced. With courses now operating within a modular and semesterized system, the luxury of design weeks has disappeared completely. Nevertheless, the heat exchanger project still continues as an important part of the design teaching, with the specification having been modified to take into account the limited time available for construction.

STUDENT ASSIGNMENT It must be stressed that the following design brief only applies to one academic year, that being the year in which the heat exchanger, instead of being part of the general design course, formed the major part of a semester-long module entitled Design of thermal systems. The module was based on two formal contact hours and one studio hour each week over a twelve-week period. Design brief Students will work in groups of four. Each group is to design a cross-flow heat exchanger incorporating several staggered rows of circular tubes, to cool water flowing at a rate of 1.5 litres min from 70C to 60C. Only one type of aluminium tube of nominally 8 mm outside diameter is available. The heat exchanger configuration will have a maximum of two waterside passes and be designed for MINIMUM tube length. Each heat exchanger will be tested in the Thermofluids laboratory. Air flow will be provided at ambient conditions within the laboratory by means of a constant-speed fan having an outlet duct of 160 mm wide by 160 mm high.
International Journal of Mechanical Engineering Education Vol 27 No 3

Design of a heat exchanger

211

Each group will design the necessary tube plates and water headers. The tube plates to be manufactured using 6 mm thick by 75 mm wide aluminium plate to hold the tubes at a predetermined pitch in a staggered formation. The headers to be manufactured using 6 mm and 12 mm perspex sheet. The header to be designed to ensure there is no leakage of water from the tubes and headers, and to incorporate water inlet and outlet pipes of 20 mm nominal diameter. Work required Formal testing of the heat exchanger will take place during the final week of the semester. Each group will attend with their heat exchanger and a transition duct, to connect the heat exchanger to the fan outlet, and be allowed 30 minutes to assemble the heat exchanger and transition duct on the rig. Each heat exchanger will be subjected to five minutes steady-state testing during which the water flow rate, inlet and outlet temperatures will be measured. Each group will submit a completed and tested heat exchanger together with a BRIEF group report to incorporate a discussion of the design procedure, comparison of the estimated and actual performance and outline how the design could have been improved on the basis of the known performance. Marking scheme Thermal design Detail design and construction Performance: matching specification optimization Presentation 20 15 30 25 10

DISCUSSION Before going on to consider the heat transfer analysis it is necessary to discuss some of the points arising from the assignment as defined above. In order to run the project on an annual basis, the specification is changed each year. One of the main changes has been the size and shape of the tubes. Over the years different circular aluminium tubes with nominal diameters ranging from 6 mm to 8 mm have been defined, interspersed with years in which rectangular brass radiator tubes have been given. These changes prevent students copying solutions from previous years. The test rig is shown in schematic form in Fig. 1. It consists of a centrifugal fan supplying air flow through a rectangular settling chamber. Since the fan is also used for other projects the design of the settling chamber is to ensure a uniform air velocity at the 160 mm square outlet. The hot water is provided by means of a header tank complete with an immersion heater. Before each test the tank is topped up and the water then brought up to temperature by means of an immersion heater. Each test run uses less than a quarter of the tank full. Using this technique it is difficult to ensure that the water enters the heat exchanger at exactly 70C, but any variation can be allowed for by calculating the effectiveness of the actual heat exchanger: E= Tw1 Tw 2 Tw1 Ta 1
International Journal of Mechanical Engineering Education Vol 27 No 3

212

K. Sherwin and M. Mavromihales

Fig. 1. Test rig.

and comparing with the design specification. Water and ambient temperatures are measured using thermometers and the water flow rate by means of a rotameter. The fan speed can be varied and, again, different settings are used each year to give an outlet air velocity from the fan of between 5 and 8 m s. During the thermal design stage students build simple models to simulate their proposed design in order to ascertain the actual fan outlet velocity using a pitot tube and inclined manometer. For the construction of the headers students are provided with material in the form of 6 mm thick aluminium plate, for the tube plates, together with 6 and 12 mm thick perspex sheets for the headers. Machining of the tube plates is carried out by the University using CAD data from the students and translating for use on a CNC machine. This means that a period of up to three weeks has to be provided between the completion of the thermal design stage and the start of the heat exchanger construction, to allow for the manufacture of all the tube plates. A certain amount of proficiency in the use of Autocad 2D CAD is assumed from earlier course modules. Some students have greater fluency in the use of CAD, often due to experience gained as part of their industrial placement. The anticipated proficiency in the use of CAD assists in further reducing the Design to Manufacture lead time whilst introducing students to best practice in CAD CAM for data integrity and for easy and ready accessibility. For example, the use of the layering system in CAD drawing practice for easy extraction of channels and other features to be machined on the tube plate is encouraged. Layers on CAD can be assigned in order to differentiate between the machining process applied. This layering system can be further applied to allocate layers in accordance to the cutting tool applied from a standard library of tools. This exposes the students to the practical constraints of manufacturing, such as tooling availability and ease of machining that govern the overall product lead time. The machine tool limitations must be noted by the students. Whilst they are encouraged to draw machined profiles to nominal dimensions, the specified finished tolerances must be consistently achievable by the available CNC machine tools.
International Journal of Mechanical Engineering Education Vol 27 No 3

Design of a heat exchanger

213

Once translated into a Data Exchange Format (DXF) file the completed drawing files are offered to University technician staff for Computer Assisted Parts Programming on a Camtek PEPS CAM system, prior to machining on the CNC machine tools. Students make their own headers using simple hand tools in a workshop provided. The specification of a single- or two-pass arrangement on the water-side reflects the need to simplify construction due to time constraints within a semester-based project. The choice of perspex sheeting for the headers provides a material that is easy to work with. Also, its low thermal conductivity ensures that only minimal heat is lost from the headers to the surroundings. Students are required to provide their own adhesives, gaskets, bolts, nuts and other minor items. This is to ensure that there are no enforced constraints on the detail assembly of the heat exchangers. The marking scheme can be varied to place particular emphasis on certain aspects of the design. Of particular interest is the fact that the majority of the marks are given for the heat exchanger performance. These are broken down into marks awarded for the heat exchanger achieving the design specification, in terms of heat transfer performance and not leaking, and marks for optimization. The latter are awarded on a linear scale with maximum marks for the heat exchanger having minimum tube length and zero optimization marks for the heat exchanger with the longest tube length. This way, a degree of business realism is incorporated in the project since, in practice, the capital cost of a heat exchanger is proportional to the tube length. It also introduces some competition into the project with students aiming for minimum length whilst, at the same time, ensuring that the heat exchanger is sufficiently large to provide the specified rate of heat transfer.

THERMAL DESIGN The thermal design of a heat exchanger provides a valuable lesson for students. Even finalyear students are surprised to find that here is a situation where the number of unknowns exceeds the number of equations available and it is necessary to use an iterative design procedure. There are two approaches to designing heat exchangers. Either the log mean temperature difference, LMTD, or effectivenessnumber of transfer units, ENTU, approach. Over the years, some student groups have preferred the latter but most have chosen the LMTD approach as being conceptually more straightforward. For this reason, the design technique described below is based on the LMTD approach. The basic equation for any heat exchanger is: Q = UATm (1)

where Tm is the mean temperature difference. Where the temperature changes within the water and air are much smaller than the overall difference in temperature between the fluids, it is sufficiently accurate to take an arithmetic mean temperature difference:
Tm = Tw1 + Tw 2 Ta 1 + Ta 2

[(

) (

)] 2

(2)

Using these equations the design can be based on an initial simple heat transfer model, in order to allow the students to rapidly get to grips with the problem, which can then be refined to make it more realistic.
International Journal of Mechanical Engineering Education Vol 27 No 3

214 First model

K. Sherwin and M. Mavromihales

The initial model can be based on three simplifying assumption: (1) (2) Thermal resistance across the wall of an aluminium tube is negligible. The heat transfer coefficient on the water-side is very much greater than that on the air-side: hw (3) ha

The temperature rise on the air-side is very small:

Ta 2 Ta 1

Based upon assumptions 1 and 2 it can be assumed that the air-side heat transfer coefficient controls the design since: U and Q ha ATm (3) ha

The temperature rise on the air-side tends not to be negligible but assumption 3 allows the initial analysis to be simplified whilst incurring errors in the rate of heat transfer of less than 10%. Using this simple model allows the students to start the design procedure and to appreciate which variables have most effect on the overall performance. In practice, the tube pitch is a key variable but students have to temper their wish to minimize the pitch with the recognition they will actually have to construct the final design. Over the years the number of rows the students have chosen has varied between two and six. In practice, there is little reason to choose a large number of rows because, although there is some enhancement of the air-side heat transfer coefficient due to increased turbulence from preceding rows, the air-side pressure drop is also increased with a resulting reduction in the air-side mass flow rate.

Second model Once students have started to understand the thermal design procedure it is necessary for them to refine their design model by including the water-side heat transfer coefficient and the temperature rise on the air-side within their calculations. The water-side Reynolds number is generally much less than 2000, implying that the flow is laminar. However, owing to the short tube length and the mixing that takes place in the headers, it can be assumed that the flow is turbulent. The water-side heat transfer coefficient can, therefore, be estimated using the DittusBoelter relationship: Nu = 0.023 Re 0.8 Pr 0.3 (4)

In practice, the water-side heat transfer coefficient is greater than that on the air-side, but not sufficiently greater as to have negligible thermal resistance. This means that, in order to minimize the tube length, it is necessary to maximize the water-side velocity by reducing the number of tubes in a single pass. This can be achieved by increasing heat exchanger height
International Journal of Mechanical Engineering Education Vol 27 No 3

Design of a heat exchanger

215

with relation to the width and increasing the number of water-side passes. However, a maximum of two was defined in the design brief so as to simplify header construction.

RESULTS The results of the heat exchanger tests are listed in Table 1. The heat exchangers were tested over two days, hence the slight differences in the ambient temperatures, and numbers represent the order of testing. Since the required duty represents an effectiveness of around 0.2, depending on the ambient temperature, it will be noted that most of the heat exchangers were over-designed in some cases by nearly 50%. Experience shows that each cohort of students approaches the project in a different way. Since they had only a limited time for testing different configurations at the thermal design stage, this cohort went for a more conservative approach relying on the heat exchangers working rather than emphasizing the optimization.

Table 1. Test results Tube length m 1 2 3 4 5 6 7 8 9 10 7.44 7.50 8.16 7.10 7.00 6.25 7.6 7.94 9.60 9.00 Water inlet C 69.5 71.5 68.5 68.0 70.5 71.5 69.5 69.0 71.0 68.0 Water outlet C 55.5 57.0 54.0 54.5 56.0 60.0 61.0 55.0 59.5

Ambient C 20 21 21 21 21 19 19 19 19 20

E 0.283 0.287 0.305 0.287 0.293 0.219 0.168 0.280 0.177

Comments

Slight leak

Best length Air lock Slight leak Major leak Air flow

In the event three heat exchangers failed to achieve the required performance. Number 7 had an air lock in one header which prevented the water flowing through all the tubes. Once they had appreciated this and installed a vent plug, the heat exchanger was satisfactory. Poor air flow distribution caused number 10 to fail. The air flow duct was hurriedly assembled at the last minute, allowing some of the air to flow past the sides of the tube bank rather than through it. Again, once this was modified the heat exchanger was satisfactory. It was impossible to test number 9 as it developed a major leak, eventually resulting in the top header splitting from the tube plate. The required rebuild could not be implemented in the time available. Although frustrating for the students involved, such failures do reinforce the need for careful detail design. The heat exchanger achieving the best length is shown in Fig. 2. Although by no means an optimum design, it is possible to achieve a minimum length of under 5 m, the students clearly gave it careful thought. This is evident in the way the number of tubes were reduced to increase the water-side velocity, by going for a longnarrow configuration.
International Journal of Mechanical Engineering Education Vol 27 No 3

216

K. Sherwin and M. Mavromihales

At the other end of the design spectrum, those heat exchangers with tube lengths of 9 and 9.6 m respectively, had a more square configuration.

Fig. 2. Heat exchanger no. 6.

OBSERVATIONS In reality, the key design feature is the pitch diameter ratio of the tubes. This needs to be minimized in order to maximize the air-side velocity between the tubes. However, this cohort of students did not have long enough at the thermal design stage to fully appreciate this point. Nevertheless, the project has proved to be successful and students learn from the actual testing of their exchanger design. Unfortunately, financial constraints mean that it is no longer possible to purchase metal tubing for this project. It is, therefore, proposed to use drinking straws as a low cost alternative form of tubing. The means that the water temperatures may need to be reduced and that the assumption of negligible thermal resistance across the tube wall is no longer valid.

REFERENCES
[1] Sherwin, K., Engineering Design for Performance, Ellis Horwood, Chichester, 1982. [2] Johnson, A. D., and Sherwin, K., A design-build-test project for first-year students, Int. J. Mech. Eng. Educ., 21(1), 2734 (1993). [3] Yates, D. A., Not another heat exchanger?, Int. J. Mech. Eng. Educ., 19(4), 273276 (1991). [4] Anderson, D. C., A project-based heat transfer course, Int. J. Mech. Eng. Educ., 20(2), 137142 (1992).

International Journal of Mechanical Engineering Education Vol 27 No 3