THERMOSIPHON EXCHANGER FOR USE IN ANIMAL SHELTERS

B.S. Larkin J.E. Turnbull R.S. Gowe Gas Dynamics Laboratory Member CSAE Animal Research Institute National Research Council Engineering Research Service Agriculture Canada Ottawa Agriculture Canada Ottawa Ottawa

INTRODUCTION The design of various types of clean ation, vapor flow, condensation and heat exchangers to give a desired heat gravity return. This transport An animal building is ventilated to recovery is well documented; the impor is much more effective than conduction; control heat, vapor, gases and odors tant question is whether design perfor the tube has an effective conductivity given off by the animals. Where the mance can be maintained under typical which is several orders of magnitude ventilation rate is determined by sets of dusty farm conditions. In the heat higher than that of a copper bar. exhaust fans controlled by interior exchanger tested, the locations where (the usual method) the heat fouling can collect are all easily acces There is a small temperature drop losses are automatically balanced to the sible, so that cleaning is simple. The tests associated with the pressure difference heat supply. With this arrangement, con evaluated the effect of fouling on the required to cause vapor flow but this is trol of excessive in the building system and ease of cleaning in a typical usually negligible. In an animal shelter is usually no problem as long as outside poultry environment. , the boiling and conden ambient temperature remains above a sing coefficients are high, or about particular minimum. However, with One problem with air-to-air heat ex 400-500o Btu/ft2h°F (2,300-2,800 colder weather a point is reached at changers is the low air-to-surface heat watts/m2°C) so that the temperature which the mean ventilation rate control transfer coefficient. For significant heat drops at the condensing and boiling sur led to balance the animal heat production transfer, large surface areas are required. faces, although not negligible, are accept is too low to remove the water vapor This heat exchanger uses standard mass- able. produced. produced finned tubing to give the required surface area economically within The advantages of this type of heat For example, in a typical well insul compact external dimensions. exchanger are: (i) Secondary surface of ated swine growing building maintained commercial finned tubing where heat is at 60° F (15°C) this heat deficit occurs at transferred between metal and air; about 25°F (-3°C) outdoor temperature. THERMOSIPHON HEAT EXCHANGER (ii) Metal/air interfaces easily accessible For a more detailed discussion see (2), for cleaning, unlike previous plate-type (5), and (6). During colder periods of a Figure 1 shows one element of a heat exchangers; (iii) Simplicity and ease typical Canadian winter the options are: thermosiphon heat exchanger. Heat is of maintenance, with no mechanical (i) to reduce ventilation rate, resulting in transferred from the exhaust to the parts, pipe connections, headers or valves high humidity, condensation and odors, inlet duct along the thermosiphon tubes. which may leak; (iv) Compact physical (ii) to maintain higher ventilation to help arrangement of the heat exchanger, fit control humidity but permit a much During assembly, the air is removed ting easily into normal rectangular duct colder building, or (iii) to add enough from the tubes. A measured quantity of ing without complicated headers or heat to maintain adequate ventilation and working fluid (in this case 22) transition pieces. temperature. is put into each tube so that the lower section is filled with liquid and the upper It is also useful to compare this heat One method of adding heat is to section contains vapor. In operation, recovery application with other buildings recover it from the exhaust air with a whenever the inlet air is colder than the such as hospitals, apartments and office heat exchanger. Giese and Downing (2) exhaust air the vapor will start to buildings. The special features of the and Giese and Ibrahim (3) studied tube condense. This reduces the pressure in the animal shelter application are: (i) The and shell heat exchangers. Giese and Bond tube and the liquid begins to boil. The exhaust air is dirty. Tolerance of fouling (1), Turnbulla and Ogilvie (4) studied resulting vapor flows to the cooler section and ease of cleaning are essential, plate-type heat exchangers. Witz and of the tube and condenses, releasing (ii) There are no specialized maintenance Prattb reported on a rock-bed heat technicians at hand. Reliability and ease storage system through which incoming latent heat. Heat is thus transported along the tube by a continuous cycle of evapor- of maintenance are more important and outgoing flows were alternated. (iii) In most buildings a heat exchanger is designed for maximum practical heat Contribution No. 468 and 545 from Engineer a Turnbull, J.E. Performance of a perpendi recovery, i.e. about 75% effectiveness. ing Research Service and Animal Research cular flow air-to-air heat exchanger. Agric. This means that the heat exchanger Institute, respectively. Eng. Ext. Release, Ontario Dep. Agric, recovers about 75% of the theoretical Guelph, Ontario, Nov. 1965. maximum heat recovery, which depends on the temperature difference between b Witz, R.L. and G.L. Pratt. Innovation in the exhaust and intake air streams. When RECEIVED FOR PUBLICATION AUGUST 6, beef confinement housing and equipment. 1974 designed to 75% effectiveness a thermo ASAE paper no. 71-208, 1971. siphon type heat exchanger is more

CANADIAN AGRICULTURAL ENGINEERING, VOL. 17 NO. 2, DECEMBER 1975 85 coefficients in a thermosiphon tube, for typical conditions in a heat exchanger. The test section consisted of thick-walled copper tube, 0.951-inch (2.41-cm) inside diameter. Vapor temperatures were measured with a thermocouple passed along a smaller tube (o.d. = 0.061 inch (1.55 mm), i.d. = 0.045 inch (1.14 mm)) running down the center of the main tube. A similar tube was buried in the wall of the main tube, so that the tube wall temperature could be measured by moving a thermocouple along its length. Using the same thermocouple for measuring all temperatures minimized calibration errors. Heat was supplied 0 REGION OF BOILING LIQUID electrically at one end and removed by a (12 OZ. OF REFRIGERANT 22) fluid flowing through a cooling jacket at 0 REGION OF CONDENSING GAS the other end. 0 ROUND SEALED TUBE OF RED BRASS, These tests led to the following empiri OUTSIDE DIAM. 1.00 IN (2.54 CM ) cal correlations, based on the inside tube INSIDE DIAM. 0.902 IN (2.29 CM ) surface: LENGTH 50 IN (127 CM ) 0 ALUMINUM FINS. H6=91 +0.07862 + 2.577 (1) THICKNESS 0.015 IN (0.38 MM ) Hc=645 - 0.0520- 1.24T (2) 6 FINS/IN (2.4/CM), PRESSED ONTO TUBE SURFACE RATIO 20.3/1 where

0 AIR FLOW DIVIDER. GALVANIZED STEEL. Hu = boiling coefficient 0 ELEMENTS SLOPED 4.5° FROM HORIZONTAL (Btu/ft2h°F) H„ = condensing heat transfer coefficient (Btu/ft2h F) Q = heat flux (Btu/ft2h)o Figure 1. Diagram of a thermosiphon tube (not to scale) with specifications as used in this T = vapor temperature ( F) experiment.

HEAT EXCHANGER

The heat exchanger was made up from thermosiphon tube elements (see Figures 1 and 2) banked vertically four tube elements per bank, with one set of aluminum fins. In the first winter nine banks of tubes were used; in the second winter this was reduced to five.

TEST OBJECTIVES

1. To demonstrate the operation of a thermosiphon heat exchanger. 1. To verify design correlations. 3. To investigate the effect of fouling on flow and heat transfer. Figure 2. Thermosiphon tube bank. 4. To study the problem of heat ex changer freezing. 5. To evaluate cleaning methods for the expensive than some other types, but at system would fill with ice. This appears heat exchanger and filters. lower effectiveness it is competitive. to be the practical upper limit on With laying chickens (a typical farm effectiveness and is typical of most farm example) 40% effectiveness is about applications. optimum; this will recover enough heat CHICKEN HOUSE INSTALLATION for good ventiliation down to -30 F (-34°C) outside temperature. Below this, LABORATORY TESTS Figure 3 shows the equipment arrange additional heat cannot come from the ment in the attic space of a research exhaust air, otherwise the exhaust flow Laboratory tests were carried out to chicken house at the Animal Research would be cooled below freezing and the measure the boiling and condensing Institute Greenbelt Farm, Ottawa.

86 CANADIAN AGRICULTURAL ENGINEERING, VOL. 17 NO. 2, DECEMBER 1975 o EXHAUST FROM CHICKEN CAGE ROOM; 4 FILTER ELEMENTS AT 16x25x2IN(40.6x63.5x5.1 CM) © EXHAUST AIR FILTER FROM CHICKEN FLOOR HOUSING ROOM; 4 FILTER ELEMENTS (SAME AS®) © CENTRIFUGAL EXHAUST , DELHI MODEL 412, 3/4 HP MOTOR © FINNED HEAT EXCHANGER BANKS IN GALV. STEEL BOX TILTED 4.5° ACROSS AND 3.5° BACK, FOR CONDENSATE DRAINAGE CONDENSATE DRAIN TUBE FROM LOW CORNER TO MANURE PIT

INSULATED EXHAUST DUCT

TRAP DOOR STATION FOR TAKING VELOCITY MEASUREMENTS. INSULATED INTAKE DUCT FROM VENTILATED ATTIC CENTRIFUGAL INTAKE FAN, SAME AS (7) DISCHARGES TO ATTIC DUCT.

NLET DUCT & ADJUSTABLE INLET HOLES TO POULTRY ROOMS

Xv^v-:->>^:-;i>x->:^^^ © ,v®

SECTION 'B-B SECTION 'A'-'A'

Figure 3. Detail of heat exchanger installation.

The ducting and equipment was fitted estimating throughputs. Chickens were direction was about 900 scfm (25.5 into a rather tight triangular duct space housed in cages in one room and on slat m3/min), about 50% higher than the within the roof trusswork, to serve two and litter floor in the other room. The recommended minimum flow. Also, the chicken rooms in the ground flow below. heat exchanger system was super small ventilating fans in the rooms ran Fresh outside air was drawn from the imposed on the conventional intake/ continuously so that the total ventilation ventilated attic space, warmed through exhaust ventilation system such that in rate was much higher than the required the heat exchanger and discharged into warmer winter weather the original minimum. the attic duct surrounding the equipment, exhaust fans would increase the ventila then into the poultry rooms below tion from the attic in steps, to control Flow temperatures and velocities were through inlets at the top of the center chicken . obtained for a range of outside tempera wall dividing the rooms. tures, using calibrated mercury thermo meters and a calibrated vane anemometer. Exhaust air was drawn through air TEST PROCEDURE filters in the ceiling of each room and January-February 1974 into the centrifugal exhaust fan, then January-February 1973 blown through the heat exchanger to Prior to this second test period the exhaust outdoors at the gable end of the The unit was operated for 21 days, nine tube banks were reduced to five. building. Straight duct sections on intake with 600 birds of average weight 3.6 lb Calculations showed that five tube banks and exhaust air flows were designed for (1.6 kg) in each room. The air flow were close to the optimum for a typical taking air velocity measurements for through the heat exchanger in each animal production unit. The heat

CANADIAN AGRICULTURAL ENGINEERING, VOL. 17 NO. 2, DECEMBER 1975 87 Results were obtained at various out side temperatures after the extra birds HEAT EXCHANGER were added to the cage room. Finally, the effect on flow rate of fouling in various parts of the system was investigated. AP--0.42 IN H20 INTAKE AIR +21.6 F -3.6 F RESULTS "(-5.8C) (_19-8C)3780 LB/HR Figure 4 shows typical performance (1750KG/HR) with five banks of thermosiphon tube elements operating.

Fouling can affect the performance of HEAT FLUX EXHAUST AIR 23,500 BTU/HR the recovery system in two ways. Firstly, ,61.2 F (6.9 KW) 39.9 F the exhaust flow is reduced. This effect **(16.2C) (4.4C) can be assessed by comparing exhaust 3410 LB/HR flows over the period of the test. Second REL.HUM.= 58% AP=0.515 IN H20 ly, the heat transfer coefficient between the exhaust air and the heat exchanger surface may be reduced. This is more complicated because both the exhaust INLET FLOW TEMPERATURE RISE: 21.6-(-3.6) =25.2 F flow and the system temperatures vary EXHAUST FLOW TEMPERATURE DROP =61.2-39.9 =21.3 F from day to day, so that a simple HEAT EXCHANGER EFFECTIVENESS =0.39 statement of the heat transferred per hour is not a valid comparison. In order to isolate this effect, an "effectiveness Figure4. Typical 5-bank heat exchanger performance diagram,for February 15,1974. factor" has been defined as the ratio of the heat transfer as measured to the heat TABLE I EXPERIMENTAL RESULTS, 1973 TESTS transfer predicted theoretically for the same temperature and flow with a clean heat exchanger. This dimensionless ratio Inlet flow Exhaust flow Effectiveness has been calculated for each set of test Day lb/h lb/h factor results as shown in Tables I (1973) and II (1974). 2 2,250 3,000 1.15 2,540 0.99 5 2,285 The significance of this parameter lies 6 2,305 2,875 1.07 in its variation during the test. Whether 7 2,195 2,980 1.15 11 2,190 3,060 1.13 the value is greater or less than unity 17 2,325 2,590 1.00 depends on whether the design correla 20 2,475 2,400 1.10 tions are conservative or optimistic and also on the errors in the experimental measurements. TABLE II EXPERIMENTAL RESULTS, 1974 TESTS January-February 1973

Inlet flow Exhaust flow Effectiveness There is considerable scatter in these Day lb/h lb/h factor results and it is not clear whether either the flow or the effectiveness factor de 1,980 1.18 17 1,770 creased during the tests. Also, much of 1,700 1,850 1.07 18 the fouling was removed from the build 21 1,870 1,540 1.14 24 1,900 1,550 1.11 ing by over-ventilation through the 25 1,890 1,705 1.08 normal ventilating system so that the dust 28 1,865 1,430 1.10 content in the part of the air flowing 32 1,745 1,245 1.13 through the heat exchanger was not 33 1,870 1,240 1.07 typical. This preliminary part of the test 34 1,815 1,352 1.11 served to demonstrate the operation of the heat exchanger, confirm the design exchanger was cleaned with a vacuum The heat exchanger was operated for relationships and give experience on the cleaner. 38 days. For the first 15 days there were freezing problem. Inspection of the tube about 500 birds (average weight 3.6 lb banks afterwards showed that most of the To control freezing of condensate in (1.6 kg)) in each room; then the number fouling was a soft fluff, which was the exhaust side of the heat exchanger in the cage room was increased to 1,080 removed with a vacuum cleaner. elements, a in the downstream birds. The electric heaters in eachpoultry exhaust duct was set to stop the intake room were set at 58°F on, 61 F off January-February 1974 fan whenever final exhaust air tempera (14.4°C, 16.1°C). The room ventilation ture approached freezing. When necessary fans were set at 67°F on, 64°F off Table II shows a reducing exhaust flow this stopped the cold intake air flow, (19.4°C, 17.8°C). The heat exchanger over the period of the test. No significant allowing uncooled exhaust air to defrost fans were set to give a flow of about 700 change in effectiveness factor was obser the tube banks. scfm(20m3/min). ved.

CANADIAN AGRICULTURAL ENGINEERING, VOL. 17 NO. 2, DECEMBER 1975 88 At the end of the run, tests were It was found that it would defrost if the house heat recovery system. carried out to determine the cause of the cold intake air flow was stopped by flow reductions. shutting off the intake fan. Dust fouling causes significant reduc tions in heat transfer rate and exhaust air First, an unsuccessful attempt was The heat exchanger froze on one flow through the heat exchanger. Further made to clean the heat exchanger with a occasion during the 1974 tests, with the tests are proposed to find a more eco vacuum cleaner. The depth of cleaning thermostat bulb positioned in the center nomical, effective and easily maintained was limited to the slight penetration of of the duct and set to switch off the filter arrangement so that the recovery of the brush between the fins; measurements intake fan at 34°F (1.1°C) exhaust the heat exchanger can be maintained. before and after showed no change in the temperature. Lowest exhaust tempera flow. tures were found in the exhaust flow nearest the inlet duct. The thermostat Next, the exhaust flow was measured bulb was moved to this location and reset with one of the filters removed, reducing to 35 F (1.7°C). No further freezing was ACKNOWLEDGMENTS the pressure drop across the filters to observed although on one occasion the zero. The resulting increase in flow was overnight minimum temperature was The contributions by H.A. Dubroy found to be only 3%. -10°Fo(-23°C) compared with -7°F and R. Alary to the design, construction, (—21.7 C) on the occasion when freezing installation and testing of the thermo Fouling also affects the occurred. siphon heat exchanger are gratefully rotor. When the rotor was cleaned the acknowledged. exhaust flow increased by 7%. DISCUSSION Finally, the heat exchanger was clean ed with a high-pressure water jet. About 5 These tests show that the heat ex minutes cleaning was sufficient to remove changer operates satisfactorily and can be REFERENCES the accumulated fouling in both intake cleaned easily. However, cleaning a heat and exhaust sides of the heat exchanger. exchanger is not as simple as cleaning a 1. Giese, H. and T.E. Bond. 1952. Design of The exhaust side was heavily fouled and filter. Therefore, a filter system should a plate-type heat exchanger. Agric. Eng. required most of the cleaning time; the extract most of the fouling before it 33:10, pp. 617-622, Oct. fouling in the inlet side was not signifi reaches the heat exchanger. The filters cant. Measurements before and after must be economical and effective enough 2. Giese, H. and C.G.E. Downing. 1950. cleaning showed that fouling had reduced to make heat exchanger cleaning infre Application of heat exchangers to dairy the exhaust flow by 28%. This was quent, and yet not block up too quickly barn ventilation. Agric. Eng. 31:4, pp. obviously the major cause of the flow themselves. 167-170, 174, April. decrease. 3. Giese, H. and A.A. Ibrahim. 1950. Ventila Neither disposable glass-fiber nor tion of animal shelters by the use of heat The fouling in the heat exchanger was washable aluminum-mesh filters were exchangers. Agric. Eng. 31:7, pp. 327-333, a layer of dust, rather than the fluff satisfactory. The disposable types were July. found after the first tests. There was too expensive to be replaced at least condensation in the tube banks during weekly. The aluminum-mesh type did not 4. Ogilvie, J.R. 1967. A heat exchanger for the second test series, which would cause remove enough fouling to keep the heat livestock shelters. Can. Agric. Eng. 9:1, dust to stick to the metal surface. exchanger acceptably clean, and the mesh pp. 31-32, 53, Jan. was too fragile. 5. Turnbull, J.E. 1973. Engineering for inten sive housing of livestock. Pub. 1503, FREEZING IN THE HEAT Can. Dep. Agric, Ottawa, Ont. EXCHANGER CONCLUSIONS 6. Turnbull, J.E. and N.A. Bird. 1971. Con During the first winter the heat The thermosiphon heat exchanger is finement swine housing. Publ. 1451, Can. exchanger froze up on several occasions. potentially suitable for use in a chicken Dep. Agric, Ottawa, Ont.

CANADIAN AGRICULTURAL ENGINEERING,VOL. 17 NO. 2, DECEMBER 1975 89