Pressure drop over the heat transfer area of a Plate Heat Exchanger (PHE) plate By Ergin Kulenovic Chemical Engineering, LTH, Lund University

Abstract The aim of this master thesis was to examine the pressure drop over the heat transfer area for three types of PHE-plates and to devise a proper method for doing so. The tests were done in a pressure drop rig designed at Alfa Laval Lund AB. A method for preparing the samples and testing them was developed. The procedure involved cutting up pieces of the heat transfer area plates and putting them together to form larger homogenic areas to be tested. This was done by water jet cutting and then point welding and silicone sealing. Three patterns were tested, A, B and C. The test results are compared with the predictions made by simulations at Alfa Laval. The results are shown in the form of a series of Fanning s friction factor Reynolds number graphs. The results differed significantly from the predicted data which leads to the conclusion that these kinds of tests are necessary for the improvement of simulation methods and the accuracy of the data given to the PHE-user. The new correlations are to be used for the tested patterns in the future and this method is to be used on future patterns. The parameter used for comparison was 1. Introduction the Fanning friction factor. PHEs are widely used within the industry 3. Method for heating, cooling, evaporating and condensation purposes which is why it is 3.1 The Samples necessary to optimize their performance. The samples tested were made out of Usage of PHEs often results in a high smaller pieces of the heat transfer area put pressure drop. In order to minimize the together in order to get a larger homogenic pressure drop and at the same time area and hence make the tests more maintain the high level of performance the reliable. pressure drop in different parts of channel This was a task that demanded some trial must be known. and error experiments. A lot of methods This was done for the heat transfer were tested for cutting the smaller pieces area by experimentally measuring the out and the method that prevailed in was pressure drop and comparing it to the water jet cutting. This was done at Water predicted one. In order to do this a pressure Jet Sweden AB. The assembly into larger drop rig was designed at Alfa Laval and areas was done by point welding the small during the course of this thesis a method pieces onto a support plate and sealing the for operating the rig was devised. connections between them (the fittings) with silicone.

2. Theory The calculations based on the obtained data were processed with the help of the correlation for the friction pressure drop: pf

(4 f

L dH

)

v2 2

(1)

3.2 The Rig The rig used in the experiments is in fact nothing but a 2000x300x30 mm channel in which plates can be inserted to simulate a passage of a medium through a PHE channel.

Figure 1: The rig with the lid off

In order to make sure that the medium only passes between the PHE plates and not around them the rig had to be sealed both vertically and horizontally. The vertical sealing was done by alternating distance sheets and rubber lining to fill out the space left between the bottom and the lid of the rig after the sample is inserted. This was not so easily done since too much filling could crush the plates after the lid is placed and too little would allow the sample channel to open when the rig is pressurized. This problem was solved by using a piece of a pressure paper , i.e. special paper that permanently changes colour depending on the pressure exerted on it. The horizontal sealing was done by gluing rubber linings on the sides of the rig. The insertion of the sample made this a little difficult since it was easy to rip the lining of with the side of the sample. This was solved by using a thicker, riffled rubber lining on one side. This lining was compressed more easily.

3.3 The Experimental Setup The setup was a simple one (fig. 3). A loop was made so that the medium tested (in this case water) could pass through the sample channel. Pressure outtake needles were inserted from the side and connected to a differential pressure meter. All of the experiments were done adiabatically. Two flow meters were used. One for the flows up to 0.7 l/s and the other one for the flows above that. The data was collected by a computer program called Easy View in a form of continuous measurements of which an average was noted. A set of 10-20 measurements (different flows which gave different pressure drops) was noted for each sample channel. The results are presented in the form of a series of Fanning s friction factor Reynolds number graphs. The measured data was compared to the predicted ones. These were obtained by simulations at Alfa Laval Lund AB.

Figure 3: Schematics for the rig setup

4. Results and discussion Three different types of plates were tested. Pattern A, B and C. 4.1 Pattern A Pattern A plate is an asymmetric plate designed to have a low pressure drop. It is mainly used for liquid and steam processes.

Figure 2: A cross-section of a packed sample (the numbers are illustratory only)

4.2 Pattern B The pattern B plates can be arranged to make three different channels: narrow (NH), wide (SH) and medium (SS). The pressure drops in these were compared to the predicted ones. The results include the comparison for the heat transfer areas as well as for the whole channels. Figure 4: The investigated angles of flow direction for pattern A. Normal flow direction being an angle of 0°

Pattern B: SH Channel

1 1000

100000

Fanning's Friction Factor

This pattern was tested for different directions of the flow since these can cause misdistribution of the medium throughout the channel. Due to lack of the sample plates only the steam side was tested. The two graphs (fig. 5 and 6) show that the predictions of the pressure drop made were too low and that the pressure drop increases with the angle of the direction of the flow.

10000

Whole channel (Measured) Whole channel (Predicted) Heat area (Measured) Heat area (Predicted)

0,1 NRe

Figure 7: The predicted and the measured values for the pattern B SH channel Pattern B: NH Channel 1 1000

Fanning's Friction Factor

Fanning's friction factor

Pattern A Steam Side 90 degrees 45

10 1 0,1 0,01 1000

10000

100000

0,1

Whole channel (Predicted) Heat area (Predicted) Whole channel (Measured) Heat area (Measured) 0,01

10000 100000 1000000 NRe

NRe

Figure 8: The predicted and the measured values for the pattern B NH channel

Figure 5: Fanning s friction factor for different inflow angles of water on the steam side of a pattern A channel

Pattern B: SS Channel 1 1000

Pattern A, Steam side: 0 degrees

10000

Whole channel (Measured) 100000

10000

100000

Fanning's Friction Factor

0 Degrees (Measured) 0 Degrees (Predicted)

0,10000

Fanning's Friction Factor

Whole channel (Predicted) 1,00000 1000

Heat area (Measured) Heat area (Predicted)

0,1

NRe

Figure 9: The predicted and the measured values for the pattern B SS channel

0,01000

NRe

Figure 6: A comparison of the predicted and measured values for the direction of the current flow

The graphs show the same difference in tendencies for the heat transfer areas as for

the whole channels. The second thing to be noted is that although they all come from the same plate the different channels show different pressure drop tendencies. 4.3 Pattern C Pattern C plates have three pressing depths within the same plate. These three were examined separately and are 5.5, 3.8 and 2.1 mm. Both the evaporation and the condensation sides were examined. The graphs for this pattern can be seen in appendix 1 because of the detail required to be seen. The predictions were that the evaporation side of the channel would be the one with the higher pressure fall while the measurements showed it to be the condensation side (App. 1, fig. I and II). The second conclusion to be drawn is that the measured pressure drops were much higher than the predicted ones and that a closer look should be taken at the evaluation methods in order to avoid these mistakes in the future (App. I, fig. III and IV). The channels seem to be very placement sensitive which means that a small movement of the plates with respect to each other could cause an increase in pressure drop. This could be a part of the explanation to why the differences between the predicted and measured values were so big for channels with small pressing depths.

5. Conclusions A method for testing the pressure drop in different parts of a PHE-plate was developed using the rig designed at Alfa Laval Lund AB. The big differences between the predicted and the measured pressure drops indicate that it is necessary to test all new patterns instead of relying on simulated predictions only. New correlations, i.e. friction factors as a function of Re have been produced for three of Alfa Laval s patterns and are to be used instead of the predictions in the future.

6. Further Work A further look should be taken at simulation methods in order to improve them. The method for testing the samples could also be improved by finding a new way to determine the pressure exerted on the sample in the rig, a softer horizontal lining and the position of the pressure outtake needles. The pressure drop over several channels instead of just one should also be examined. The pressure drop for several channels is not necessary just a multiple of one channel.

Acknowledgements I would like to thank my supervisor Dr. Matz Andersson and the rest of the R&D department at Alfa Laval Lund AB for guiding me throughout my thesis and for taking the time to answer all my questions no matter what they were. I would also like to thank Mr. Bengt Göland and Mr. Michel Granath at Alfa Laval Lund for giving me all the technical support without which there is a chance I never would have finished. Finally, I would like to express my thanks to Prof. Anders Axelsson at Chemical Engineering for all of his advice.

References Applied Thermodynamics for Engineering Technologists, 4th Ed. T.D. Eastop & A. McConkey Gasket and plate materials of Alfa Laval heat exchangers: Ladislav Novak, Alfa Laval thermal AB Värmetransport del 1 Kemisk Apparatteknik, LTH, 2001 Strömningsteknik Kemisk Apparatteknik, LTH, 1989 Conversations with: Dr. Matz Andersson Dr. Rolf Christensen Dr. Henrik Kockum Martin Holm and Leif Hallgren at Alfa Laval AB

Appendix 1 Pattern C evaporation channels Evaporation 5.5 Condensation 5.5 Evaporation 3.8 Condensation 3.8 Evaporation 2.1 Condensation 2.1

10

1

0,1 100

1 100

1000

10000

100000

100000

Evaporation 5.5 Predicted E 5.5 Evaporation 3.8 Predicted E3.8 Evaporation 2.1 Predicted E 2.1

0,01 N(Re)

Figure I: Fanning s friction factor for different measured channels in the pattern C

Figur III: Fanning s friction factor for different evaporation channels in C pattern

Pattern C (Predicted)

Pattern C condensation channels Condensation 2.1 Predicted C 2.1 Condensation 3.8 Prediction C 3.8 Condensation 5.5 Predicted C 5.5

1000

10000

100000

0,10000

Fanning's friction factor

10

1,00000

Fanning's Friction Factors

10000

0,1

NRe

100

1000

Fanning's friction Factor

Fanning's friction factor

Pattern C (Measured)

1

0,1

Condensation 5.5 Predicted Condensation 3.8 Predicted Condensation 2.1 Predicted Evaporation 5.5 Predicted Evaporation 3.8 Predicted Evaporation 2.1 Predicted

0,01000 NRe

Figure II: Fanning s friction factor for different predicted channels in the pattern C

0,01 100

1000

10000 N(Re)

Figure IV: Fanning s friction factor for condensation channels in C pattern

100000