Purging and Air Removal Richard A. Beier Mechanical Engineering Technology Department Oklahoma State University Stillwater, OK 74074 rick.beier@okstate.edu Introduction Ground source heat pump systems (GSHP) often use closed loops to couple the ground with the heat pump as illustrated in Figure 1. Plastic pipe may be installed in vertical boreholes, horizontal trenches, or bodies of water. The chosen configuration for a particular application depends on the amount of available land area, underground soil/rock structure and availability of surface water. In these systems water (or water/antifreeze mixture) circulates through the plastic-pipe loops and exchanges heat with the refrigerant through a heat exchanger at the heat pump. Then the water flows through the loops to exchange heat with the ground (or surface water). In this way heat is transferred between the heat pump and ground (or surface water). Prior to operation, the closed loops must be filled with water (or water/antifreeze mixture) and all the air in the loop must be removed. If a significant amount of air remains in the loops, problems are likely to occur in the subsequent operation of the GSHP. Air inside the loop may shorten the life of the circulating pump, which is cooled by the circulating water. Usually multiple ground loops are connected together in a header or manifold (Figure 2). The header is designed to distribute the water flow evenly among the loops. Air pockets can lead to uneven water flow among the loops. In fact, an air pocket can restrict water flow and even block the flow of water to one or more the loops, causing them to be inactive. Uneven distribution of water flow rates among the loops reduces the heat transfer capacity of the ground loops. Thus, removing air within a loop system is very important. Pressure Loss Created By Air Pockets In a ground loop air pockets may form at locations of high elevation where bubbles coalesce. The term air pocket refers to a larger volume of air that does not move with the liquid flow but remains in one location. The term air bubble refers to a smaller volume of air that moves with the liquid. Any air pocket creates a relatively large pressure drop and can alter the liquid flow distribution through a multi-loop system. Lubbers (2007) measured the additional pressure loss in water flow created due to an air pocket. He studied the effects of an air pocket in a downward sloping pipe. In the experiments both water and air are supplied upstream of the air pocket to maintain the size of the air pocket. Although his geometry is different from a ground loop, the same 1 8/17/2009 mechanisms are expected to carry over to the ground loop geometry. Figure 3 shows one flow pattern observed by Lubbers where a stationary air pocket is located in the sloping pipe. At the lower end of the air pocket a hydraulic jump may exist between the air pocket and the water-filled pipe. The mixing motion in the hydraulic jump entrains air and forms smaller bubbles from the air pocket. These small bubbles move downstream. In the purging process of a ground loop the objective is to eliminate such air pockets. But in Lubbers’ experiment large pockets of air do not move. Instead, air moves as smaller bubbles break off from the pocket and the bubbles are swept downstream. This process of the air pocket shedding bubbles may take a substantial amount of time to eliminate the air pocket. Lubbers (2007) measured the additional pressure drop in the liquid flow caused by the stationary air pocket. Figure 4 illustrates the total pressure drop (vertical axis) across the downward sloping pipe. For these data the slope of the pipe is 10 degrees from the horizontal plane. The water flow rate is on the horizontal axis (ft/sec). The pressure drop for a liquid filled pipe (no air) is shown by the solid curve at the bottom of the graph. The gray band represents the spread of data over a range of air flow rates. Note the large increase in the pressure drop when a gas pocket is present, especially at low liquid flow rates. The air pocket (Figure 3) is present for liquid flow rates between 0.7 and 4.3 ft/sec. The pressure loss generally decreases as the average water rate increases. Also, the pressure drop decreases as the air flow rate decreases. These trends correspond to larger air pockets producing larger pressure drops. Lubbers also studied the flow patterns to sweep air down a vertical pipe. The pressure drop curve for a vertical pipe in Figure 5 shows the pressure drop decreases sharply with increasing liquid velocity. This decrease is more abrupt than the decreases observed in pipes with slope angles of 5 degrees to 30 degrees from the horizontal plane. Again, the gray band represents the spread of data over a range of air flow rates. Lubber argues that bubbles with diameters of 0.1 to 0.2 inches rise at velocity of about 0.7 to 0.8 ft/sec in a still water column. The results in Figure 5 indicate a downward liquid velocity greater than 0.8 ft/sec tends to sweep the bubbles downstream in a vertical pipe. Thus, air bubbles are more readily swept down a vertical pipe than a sloping pipe. In a pipe where the angle from the horizontal is less than 90 degrees, the bubbles have a tendency to travel in the upper part of the cross section of the pipe with bubble-free liquid flow in the bottom region. Then the bubbles are more likely to coalesce and form a stationary gas pocket. A larger liquid flow rate is needed to carry the air downstream compared with the case of a vertical pipe. At moderate liquid flow velocities Lubbers’ experiments demonstrate the transport of air occurs by sweeping small bubbles downstream. Larger air pockets must be broken down into smaller bubbles in order to eliminate the gas pockets. The breakdown process takes a significant amount of time to complete. 2 8/17/2009 Size of Air Bubbles The transport of air in water pipelines has been studied in applications of civil engineering. Escarameia (2005) and Lubbers (2007) have recently reviewed the technical information on the movement of air in pipes and pipelines. Drag and buoyancy forces are exerted on an air bubble in flowing water. The drag force is proportional to the projected cross-sectional area of the bubble. The buoyancy force is proportional to the volume of the bubble. A convenient ratio is the drag force on a bubble over the buoyancy force on the bubble. The ratio of forces is related to the bubble diameter by drag force d2 ∝ 3 buoyancy force d (1) As the diameter decreases, the drag force pushing the bubble downstream increases relative to the buoyancy force. Thus, smaller bubbles are more easily swept downstream. Researchers (Levich, 1962; Hinze, 1955) have argued that there is a maximum stable bubble diameter in turbulent liquid flow through a horizontal pipe. In their analysis the air fraction is small enough so that bubbles do not coalesce. Pressure fluctuations in turbulent flow tend to change the shape and breakup a bubbles. On the other hand, surface tension forces tend to resist any deformation of the bubble. Based on these principles Hesketh et al. (1987) have developed an equation to estimate the largest stable bubble size for air being carried by in turbulent flow. Removal Velocity for Air Pocket Movement Of particular interest is the minimum water velocity to move air through a pipe system. Researchers have carried out experiments to determine the minimum velocity needed to move air bubbles downward through a sloping pipe. Usually the experimental results are written in terms of a Froude number (Ervine, 1998) Fr = Uc (2) gD where Uc = minimum or critical water velocity for air pocket movement g = gravitational constant D = pipe diameter The dimensionless Froude number represents the ratio of inertia forces to gravitational forces. Experiments suggest the value of the critical Froude number for air movement depends on the downward slope of the pipe, which is represented by the angle, θ, with respect to a 3 8/17/2009 horizontal line. Ervine (1998) reviewed the experimental work by others and plotted the data on a single graph. A summary of data from Gardenberger (1957), Kent (1952), Kalinske and Robertson (1943) and Martin (1976) is illustrated in Figure 6. The results suggest the removal velocity (through the Froude number) increases with increasing pipe slope until the angle becomes 50 to 60 degrees. The data from Gardenberger (1957) indicate the required removal velocity decreases at steeper angles. The graph suggests a Froude number as large as one is needed to move air pockets through pipes. For a one-inch nominal SDR-11 HDPE pipe with an inside diameter of 1.077 inches the critical velocity is U c = Fr g D = (1.0) (32.2ft / sec 2 )(1.077 in ) 12 in / ft (3) U c = 2 ft / sec Note the result is written with only one significant digit, because the scatter among available data sets suggests significant uncertainty. The results for other pipe sizes are listed in Table 1 for pipes with slopes of 50 degrees and 0 degrees (horizontal pipes). The removal velocity increases to 3 ft/sec for a 4-inch pipe with a 50-degree slope. These removal velocities correspond to the peak Froude number in Figure 6. If the pipe is nearly horizontal, the removal velocity according to Figure 6 corresponds to a Froude number of about 0.6. The dependence of the removal velocity on the slope of the pipe, θ, has been studied by several researchers. Escarameia (2005) recommends the equation Uc = b + 0.56 (sin θ) 0.5 gD (4) The equation may be valid for angles up to 40 degrees. The value of b ranges from 0.45 to 0.61 based on the volume of the air pocket relative to the diameter of the pipe. In practice one does not know the size of the air pockets to be removed. Thus, a value of 0.6 for a can be used. Wisner et al. (1975) write their expression for the critical velocity as Uc = 0.825 + 0.25 (sin θ) 0.5 gD (5) The relation between removal velocity and slope are the same in Equations 4 and 5. Both expressions suggest a Froude number of about 1 is needed to move air through a pipe with a slope of 40 degrees. More information about removal velocity is given in reviews by Lubbers (2007) and Escarameia (2007). 4 8/17/2009 Velocity in Ground-Loop Systems The design of the piping layout for a set of boreholes should take into account the need for purging the system of air. Shut-off valves and purging connections need to be located to allow one to purge individual sections of the system. From the discussion above, sufficient water velocity is needed in the purging process to remove air from the system. For discussion purposes water flow through a header is analyzed here. A header (or manifold) is used to supply water to multiple ground loops in a parallel arrangement. Figure 7 illustrates the design for a 10-ton header, which supplies water to 10 wells, each with a depth of 200 feet. Bose (1988) discusses this header design and others. Supply water enters from the left side and enters the branch line to each borehole numbered 1 through 10. The supply pipe diameter is reduced as the water flow rate decreases in the supply header from left to right. The flow rates and pressure drops through the header during the purging process can be analyzed by using one of the many commercial software packages for pipe networks. For illustrative purposes the software program AFT Fathom® from Applied Flow Technology is used here, but other software programs can be used. The pipe network in Figure 8 is illustrated with the graphical interface inside of AFT Fathom®. The network represents the supply header (Figure 7), ground loops and return header. The water supply (40 gal/min) in the upper left-hand corner flows through the supply header across the top. The ground loops are in the middle of the diagram, and the return header is at the bottom. In the return header, 3/4 - inch pipe is on the left side and 2-inch pipe is on the right side. The U-bend at the bottom of each ground loop is represented by two 90degree elbows. As shown in Figure 9, the distribution of water velocity among the ground loops falls between 2.0 and 2.4 ft/sec. Thus, a reasonably uniform flow distribution is achieved among the loops. The velocity along the length of the header is shown in Figure 10. The downward steps correspond to take-off pipes to ground loops. The upward steps are associated with a reducer in the header. Note the lowest velocity of 2 ft/sec occurs in the 1-1/2 inch and 3/4 inch pipes, which should be sufficient to purge air from the system. Figure 11 is a similar graph for the velocity in the return header. Static pressures along the supply and return headers are shown in Figure 12. The pressure drop across the entire system (9 psi) is the difference between the pressure at the entrance of the supply header (34 psig) and the pressure at the exit of the return header (25 psig). 5 8/17/2009 Dissolved Air in Water Even if air bubbles are not present in the ground loops, some air is present in the loop as dissolved air in water (or anti-freeze solution). Consider an uncovered glass of water in a room. The interface between the liquid water and the air in the room is obvious, but the interface is not an impermeable wall separating all water molecules from air molecules. From our experience we know some liquid water may evaporate and appear as water vapor in the air. Terms such as relative humidity are used to quantify the amount of water vapor in air. On the other hand, some air is also dissolved in the water. The volumetric concentration (Lu and Likos, 2004) is one way of quantifying the amount of air in solution with water. The volumetric concentration is expressed as Va = ha Vl (6) where Va = volume of dissolved air Vl = volume of liquid ha = volumetric coefficient of solubility (volume/volume) At 68 ºF and one atmosphere of pressure ha is approximately 0.0187. That is, if the air is taken out of solution, the volume of air originally dissolved in water can be as large as 1.87 percent of the volume of water. Consider a water-filled U-tube inside vertical borehole with a depth of 300 feet. If all the air is taken out of solution from the water the air could occupy as much as 1.87 percent of the U-tube length or 5.7 feet. Thus the amount of dissolved air can be significant. Keep in mind the numbers written here are for the maximum air that can be dissolved in water. Some of this air will nearly always remain in solution. The volumetric coefficient of solubility depends on temperature and pressure as shown in Figure 13. The effects due to temperature changes are of particular interest in ground loops. If the water temperature increases from 68 ºF to 104 ºF the coefficient of solubility decreases from 0.0187 to 0.0141. Thus if the U-tube (depth of 300 ft) is filled with water at 68 ºF and experiences this temperature increase, the air coming out of solution could occupy as much as 0.46% percent of the length or 1.4 ft. Of course, if air comes out of solution, the air will form in small bubbles much like carbon dioxide gas appearing after a bottle of soda is opened. The carbon dioxide comes out of the soda as the pressure is decreased on opening the bottle. This demonstrates that the volumetric coefficient of solubility decreases with decreasing pressure. The same is true for air as seen in Figure 13. 6 8/17/2009 Addition of Antifreeze Once the air is purged from the system, the purging equipment is used to add antifreeze. When adding antifreeze to the purging tank, the antifreeze should enter the tank below the water line. In this way less air is entrained with the added antifreeze. Another issue is foaming when methanol is used as the antifreeze. Chen et al. (2007) followed an ASTM test standard (D3601-88) to assess the foaming tendency of methanol / water mixtures. In this test procedure a mixture is placed in a bottle and shaken. Foam heights are then measured. For a methanol concentration of 30% by weight, Chen et al. observed foams that lasted for 20 to 25 seconds. For methanol concentrations of 50% or greater by weight, they observed no foaming. 7 8/17/2009 References ASTM Test Standard, D3601-88 (re-approved 1997). Bose, J. E. 1988. Closed-Loop/Ground-Source Heat pump Systems: Installation Guide, International Ground Source Heat Pump Association, Stillwater, OK. Chen, G. X., Cai, T. J., Chuang, K. t., and Afacan, A., 2007, “Foaming Effect on Random Packing Performance,” Chemical Engineering Research and Design, Transactions IChemE, Vol. 85 (A2), pp. 278-282. Ervine, D. A., 1998. “Air Entrainment in hydraulic Structures: A Review,” Proceedings of the Institution of Civil Engineers. Water, Maritime and Energy, Vol. 130, pp. 142-153. Escarameia, M. (Editor), 2005. Air Problems in Pipelines: A Design Manual, HR Wallingford, Ltd., Howbery Park, Wallingford, Oxfordshire, UK. Escarameia, M., 2007. “Investigating Hydraulic Removal of Air From Water Pipelines,” Proceedings of the Institution of Civil Engineers. Water Management, Vol. 160, No. 1, pp. 25-34. Gardenberger, W., 1957. Uber die wirtschaftliche und Betriebssichere gestaltung von fernwasserleitungen. R. Oldenburg Verlag, Munich. Hesketh, R. P., Russell, T. W. F., Etchells, A. W., 1987. “Bubble Size in Horizontal Pipelines,” AIChE J, Vol. 33, pp. 663-667. Hinze, J. O., 1955. “Fundamentals of the Hydrodynamic Mechanism of Splitting in Dispersion Processes,” AIChE Journal, Vol. 1, p. 289. Kalinske, A. A. and Robertson, J. M., 1943, “Closed Conduit Flow,” Transactions of the American Society of Civil Engineers, Vol. 108, pp. 1435-1516. Kent, J. C., 1952. The Entrainment of Air in Water Flowing Through Circular Conducts with Downgrade Slopes, PhD Thesis, University of California, Berkeley, 1952. Levich, V. G., 1962. Physiochemical Hydrodynamics, Prentice Hall, Englewood Cliffs, NJ. Lu, L. and Likos, W. J., 2004. Unsaturated Soil Mechanics, John Wiley & Sons, Hoboken, NJ. Lubbers, C. L., 2007. On Gas Pockets in Wastewater Pressure Mains and Their Effect on Hydraulic Performance, IOS Press, Deft University Press, The Netherlands. 8 8/17/2009 Martin, C. S., 1976. “Vertically Downward Two-Phase Slug flow,” Transactions of the ASME Journal of Fluids Engineering, Vol. 98, No. 4, pp. 715-721. Razzaque, M. M., Afacan, A., Liu, S., Nandakumar, K., Masliyah, J. H., and Sanders, R. S., “Bubble Size in Coalescence Dominant Regime of Turbulent Air-Flow Through Horizontal Pipes,” International Journal of Multiphase Flow, Vol. 29, pp. 1451-1471. Wisner, P. E., Mohsen, F. N. and Kouwen, N., 1975. “Removal of Air From Water Lines by Hydraulic Means,” Proceedings of the American Society of Civil Engineers, Journal of the Hydraulics Division, Vol. 101, No. HY2, pp. 243-257. 9 8/17/2009 Table 1. Critical water velocity to remove air from HDPE (SDR-11) pipe of various slopes and diameters. Slope From Horizontal (degrees) Nominal Size (in) Inside Diameter (in) Water Velocity (ft/sec) Volume Flow Rate (gpm) 0 0 0 0 0 0 0 50 50 50 50 50 50 50 3/4 1 1-1/4 1-1/2 2 4 6 3/4 1 1-1/4 1-1/2 2 4 6 0.860 1.077 1.358 1.554 1.943 3.682 5.421 0.860 1.077 1.358 1.554 1.943 3.682 5.421 0.9 1.0 1.1 1.2 1.4 1.9 2.3 1.5 1.7 1.9 2.0 2.3 3.1 3.8 1.7 2.9 5.2 7.2 13 60 160 2.8 4.8 8.6 12 21 100 270 10 8/17/2009 CL/GS System Configuration (Typical Residential) Pumps and Service valves located externally Pump Module may be located inside heat pump cabinet Figure 1. Closed loop configuration (Bose, 1988). 11 8/17/2009 Figure 2. Header configurations (Bose, 1988). Figure 3. Flow pattern in sloping pipe (Lubbers, 2007). 12 8/17/2009 Pressure Loss (ft of water) 3.5 3 2.5 2 Increasing Fraction of Air 1.5 1 0.5 No Air 0 0 1 2 3 4 5 6 Velocity of Water (ft/sec) Figure 4. Pressure (head) loss in 10-degree sloping pipe over ranges of water and air flow rates (Lubbers, 2007). Diameter of pipe is 8.7 inches. 13 8/17/2009 Pressure Loss (ft of water) 8 6 4 Increasing Fraction of Air 2 0 0 1 2 3 4 5 6 Velocity of Water (ft/sec) Figure 5. Pressure (head) loss in vertical pipe over ranges of water and air flow rates (Lubbers, 2007). Diameter of pipe is 8.7 inches. 14 8/17/2009 1.2 1 Froude Number, 0.8 Garenberger (1957) 0.6 Kent (1952) 0.4 Kalinske & Robertson (1943) Martin (1976) 0.2 Kalinske & Martin 0 0 20 40 60 80 100 Angle From Horizontal (degrees) Figure 6. Velocity, Uc, required to move air down sloping pipe (Ervine, 1998). 15 8/17/2009 16 8/17/2009 Figure 7. Supply / return header for a 10-ton system. 17 8/17/2009 J1 v P1 P20 P15 P65 J57 P22 P66 J46 P21 P27 J58 P67 P33 J47 P5 J59 J48 P68 P69 P6 P38 J50 P37 J34 P36 J33 P35 J61 P7 P70 J17 P34 J7 Return Header J60 P32 J49 P31 J32 P30 J31 P29 J16 P28 J6 Supply Header Ground Loops P26 J30 P25 P4 J5 P8 J62 J63 P48 J52 J38 P46 P72 J64 P73 P11 J65 P74 P54 J53 J20 P49 J11 P53 J40 P52 J39 J22 P50 P10 J37 J10 P51 J51 P43 P9 P45 P47 P71 J19 P44 J9 P42 J36 P41 J35 P40 J18 P39 J8 P12 J66 P75 P59 J54 P58 J42 P57 J41 P56 J21 P55 J12 Figure 8. Pipe network for modeling supply header, ground loops and return header. J56 P17 J45 P16 J28 J27 J25 J26 P24 P19 P14 J29 J15 J14 P23 J13 P3 J4 P18 P2 J3 P13 J2 J67 P76 P64 J55 P63 J44 P62 J43 P61 J24 P60 J23 J68 Velocity (ft/sec) 2.4 2.2 2 1.8 10 9 8 7 6 5 4 3 2 1 Ground Loop Number Figure 9. Distribution of water velocity through ground loops. 18 8/17/2009 Velocity (ft/sec) 8 6 4 2 0 0 50 100 150 200 Flow Length (ft) Figure 10. Water velocity along the length of the supply header. 19 8/17/2009 Velocity (ft/sec) 8 6 4 2 0 0 50 100 150 200 Flow Length (ft) Figure 11. Water velocity along the length of the return header. 20 8/17/2009 Static Pressure (psig) 40 35 Supply Header 30 Flow Direction Return Header 25 20 0 50 100 150 200 Flow Length (ft) Figure 12. Static pressure along the length of the supply and return headers. 21 8/17/2009 Volumetric Coefficient of Solubility 0.14 0.12 0.1 60 psig 0.08 30 psig 0.06 0.04 0 psig 0.02 0 0 50 100 150 200 250 Temperature (ºF) Figure 13. Volumetric coefficient of solubility of air in water. 22 8/17/2009