BY THOMAS M. WALSKI, BRIAN LUBENOW, AND JEFFREY SPAIDE When they install a branch from a water distribution main, water utility managers often debate the benefits of using a tap as opposed to shutting down the line to install a T- fitting. Part of the discussion centers around the fact that a tap, especially a size-on-size tap, will have a smaller orifice size than will a T-fitting. The concern is that this will lead to additional head loss and decreased capacity. For low-flow conditions, this loss in capacity is negligible, but during fire flows, this loss can become significant. This article reports the results of experiments that indicate the loss of carrying capacity in distribution systems is not significant as long as the orifice (tap cutout) is greater than 90% of the pipe size. Head Loss IN TAPPING SLEEVES W hen a water utility needs to tap into a pipe that is already in service, there are usually two approaches: (1) shutting down the water line and cutting in a T-fitting or (2) tapping into the line under pressure using a tapping sleeve and gate. Usually tapping into the line under pressure is the preferred approach because the line being tapped does not need to be shut down. However, the head loss through the tap is usually greater than that in a T-fitting because the tap WALSKI ET AL PEER-REVIEWED 94:1 JOURNAL AWWA JANUARY 2002 91
Head loss is read at the manometer. 150 mm (6 in.) hydrant branch off a 150 mm (6 in.) main water line. Although the minor loss coefficient (K) for head loss through the branch of a T-fitting is well documented (Karney, 2000; Idelchik, 1999; Roberson & Crowe, 1997; Walski, 1984; Hydraulic Institute, 1979; Miller, 1978; Villemonte, 1977; Crane Company, 1969; and ASCE, 1965), the authors were unable to find any values for the loss coefficient of a tapping sleeve in any previous studies. This article describes experimental work to determine minor loss coefficients for tapping sleeves and analyzes the decrease in flow capacity of a tapping sleeve versus a T- fitting. LABORATORY EXPERIMENTS The experimental portion of the work consisted of measuring the velocity and head loss through a range of cutout hole sizes for taps and T-fittings. It also calculated minor loss coefficients (K) using hole is smaller than the internal pipe diameter and the tap is not as smooth as a T-fitting. The extra head loss in the tap is most pronounced in the case of size-on-size taps, such as 150 by 150 mm (6 by 6 in.) because the opening in the pipe being tapped cannot be the full diameter, because this would weaken the pipe. Manufacturers of tapping sleeves recommend that the cutout be at least 6 mm (0.25 FIGURE 1 in.) less than the full diameter of the pipe being tapped. Under normal demand conditions, A the difference in head loss between a T-fitting and a tap is considered to be negligible. However, as the velocity of water flowing through the branch increases, the head loss increases and the difference in minor loss coefficient may become significant. The most common occurrence of high velocity and head loss in the branch run of a T-fitting is that of a K = 2 g h/v 2 (1) in which K is the minor loss coefficient; h is the head loss, ft (m); and V is the velocity, fps (m/s). The apparatus for conducting the experiments was set up in the Wilkes University laboratory and is shown in the photographs. It consisted of a sump, submersible pumps, rotameter flowmeters, a test section with T-fittings and taps, throttling valves, and either a manometer or differential pressure gauge for measuring Definition of one- and two-directional flow B 92 JANUARY 2002 JOURNAL AWWA 94:1 PEER-REVIEWED WALSKI ET AL
head loss across the fitting. The apparatus was made up of 375 mm polyvinyl chloride pipe. Actual diameters rather than nominal diameters were measured and used in all calculations. The apparatus was set up so that head loss could be measured for flow entering the branch of a T-fitting with flow coming from one direction or two directions, as shown in Figure 1. All of the published values that the authors could find for K for the branch of T-fittings were only taken for onedirectional flow, yet frequently flow can approach from both directions. Each experiment described in this article was run for both one- and twodirectional flow. For each experiment, the velocity was varied by roughly a factor of three from the lowest to the highest velocity. As velocity became too low, it became difficult to read the head loss accurately. Approximately 20 data points were taken for both the oneand two-directional flow cases. EXPERIMENTAL RESULTS Tests were run for the following conditions for one- and two-directional flow. In Table 1, D refers to the diameter of the branch run, and d refers to the diameter of the tapping hole. To check the consistency of results, a plot of V 2 /2g versus h was prepared for each case. A typical curve is shown in Figure 2. This comparison showed that K was constant over the range of velocities studied, which proved that Eq 1 was valid. The runs also showed that the differences between the one-directional and two-directional flows were negligible, so that the same K could be used for either case. The value of K of 1.8 for a T-fitting was consistent with values reported in the literature for that type of fitting, indicating that the experimental procedure was correct. The key question in installing a tap was the size of the cutout (orifice) in the pipe being tapped. To cover numerous possibilities, a wide range of tapping holes was used, and it was found that the head loss was FIGURE 2 Head Loss ft (m) 1.00 (0.30) 0.90 (0.27) 0.80 (0.24) 0.70 (0.21) 0.60 (0.18) 0.50 (0.15) 0.40 (0.12) 0.30 (0.09) 0.20 (0.06) 0.10 (0.03) 0.00 FIGURE 3 Minor Loss K 1,000 10 1 TABLE 1 Typical head loss versus velocity head experiment curve 0.00 0.05 (0.01) One-directional flow Two-directional flow 0.10 (0.03) 0.15 (0.04) y = 1.86x n = 30 R 2 = 0.99 F sig = 3.31 E 33 0.20 (0.06) Minor loss versus size of orifice One-directional flow Two-directional flow Orifice y = 1.81x n = 34 R 2 = 0.99 F sig = 1.53 E 37 0.25 (0.07) V 2 /2g ft (m) 0.30 (0.09) 0.35 (0.03) 0.40 (0.12) 0.45 (0.13) 0 0.2 0.4 0.6 0.8 1.0 Diameter ratios for taps used Condition Standard T-fitting 1.000 Tap 0.941 Tap 0.784 Tap 0.627 Tap 0.471 Tap 0.314 0.50 (0.15) WALSKI ET AL PEER-REVIEWED 94:1 JOURNAL AWWA JANUARY 2002 93
FIGURE 4 Equivalent pipe length versus size of orifice with an R-value (correlation coefficient) of 0.987,000 One-directional flow Two-directional flow or K = 1.97 () 4 (3) L/D 10,000 1,000 with an R-value of 0.998. Another issue was that a tapping gate valve was not used at the tap or T- fitting. However, a gate valve would normally be used with either a T-fitting or a tap. Therefore, it would be expected that the head loss of the gate valve would be roughly the same for the tap and the T-fitting. Gate valves were not included in the experimental setups. The open gate valve minor loss coefficient of 0.39 was fairly insignificant compared with the tap orifice losses. FIGURE 5 Flow gpm (L/s) 1,800 (113.5) 1,600 (.9) 1,400 (88.3) 1,200 (75.7) 1,000 (63.0) 800 (50.4) 600 (37.8) 400 (25.2) 200 (12.6) 0 0 0.2 0.4 0.6 0.8 1.0 Flow capacity in strong system for range of available head and orifice sizes 0 20 (6) 40 (12.1) 60 (18.2) 80 (24.3) = 1 = 0.941 = 0.784 = 0.627 = 0.471 = 0.314 (30.4) 120 (36.5) Head Available ft (m) 140 (42.6) highly dependent on the hole size. The primary result of this article is shown in Figure 3, which illustrates the relationship between the minor loss coefficient (K) and the ratio of opening size to pipe diameter (). In addition to the graphical relationship between K and, a regression analysis showed that K could be determined using K = 1.28 exp (6.92 [1 ]) (2) 160 (48.7) 180 (54.8) 200 (60.9) DISCUSSION OF RESULTS One question that could be asked about the results is how they compared with the minor loss coefficient for an orifice with the same value of. The line for the orifice loss coefficient from Miller (1978), plus a minor loss coefficient of 2 for the bend, is shown as the dashed line in Figure 3. The study indicated that the tap and bend had a head loss comparable to the sum of the minor losses from an orifice in series with a bend. One obvious shortcoming of the experiment was that it was conducted in small-size piping, whereas real systems use larger piping (e.g., at least 150 mm [6 in.] for fire hydrant laterals). Because the results were developed in dimensionless form and the Reynolds numbers encountered during the experiment were in the fully turbulent range, the results are likely to be valid for larger pipes. However, additional experimental work to verify this should be conducted. IMPLICATIONS FOR DESIGN Although numerical values for minor loss coefficients are needed for modeling, it is sometimes better for design purposes to think of the minor loss in terms of the length of equivalent pipe that would cause the same head loss. Figure 4 shows the results of the experiments in terms of equivalent length of pipe based on a Darcy Weisbach f of 0.02. For example, for a of 0.8, the value of L/D is 94 JANUARY 2002 JOURNAL AWWA 94:1 PEER-REVIEWED WALSKI ET AL
This experimental setup shows a tap in the foreground and a T-fitting in the background. 220. Therefore, for a 150 mm (6 in.) pipe, the loss caused by a 120 mm (4.8 in.) tap would be equivalent to the loss caused by 34 m (110 ft) of pipe. The real question for design was the magnitude of the reduction in flow from fire hydrants when a tap was used in place of a T-fitting, especially in the case of a 150 mm (6 in.) lateral line coming off a 150 mm (6 in.) main line. There is not a straightforward answer to this, because the answer depends on the relative magnitude of the head loss in the fitting compared with the head loss in the remainder of the system. FIGURE 6 Flow gpm (L/s) 700 (44.1) 600 (37.8) 500 (31.5) 400 (25.2) 300 (18.9) 200 (21.6) (6.3) 0 Flow capacity in weak system for range of available head and orifice sizes 0 20 (6) = 1 = 0.941 = 0.784 = 0.627 = 0.471 = 0.314 40 (12.1) 60 (18.2) 80 (24.3) (30.4) 120 (36.5) Head Available ft (m) 140 (42.6) If the system is composed of long runs of small-diameter pipe, there will be a great deal of head loss between the source and the fitting, and the impact of the fitting will be small. On the other hand, if the fitting is close to an elevated water tank and is tied in to large piping, there will be relatively little 160 (48.7) 180 (54.8) 200 (60.9) head loss in the distribution grid, and the head loss in the fitting will be larger by comparison. In a specific situation, the best way to determine the impact of a tap versus a T-fitting is to model the flows for that particular situation. For this article, some calculations were made for two typical situations: a strong system in which the fitting was near a 600 mm (24 in.) pipe close to a tank and a weak system in which the fitting was connected to the tank through a long run of 150 mm (6 in.) pipe. These two conditions were simulated using a model* for a range of available heads, in which available *WaterCAD, Haestad Methods Inc., Waterbury, Conn. WALSKI ET AL PEER-REVIEWED 94:1 JOURNAL AWWA JANUARY 2002 95
FIGURE 7 70 60 Loss of hydraulic capacity as function of tap orifice size Weak 6 in. (150 mm) Strong 24 in. (600 mm) values of as shown in Figure 7. The implication was that if was at least 0.8, which was usually the case, the reduction in hydrant flow was less than 5%. Because most cutouts for mains being tapped are greater than 0.8 (), the use of a tap as opposed to a T-fitting will not significantly affect flow. Percent Reduction in Flow 50 40 30 20 10 SUMMARY This article reported the relationship between minor loss coefficients and size of the tap over a range of tap hole sizes. Although a tapping sleeve will introduce more head loss than a T-fitting, the loss in capacity is usually not significant for most applications. 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 heads is defined as the difference between water level in the tank and elevation of the hydrant. The value calculated as the indicator of flow capacity was the free discharge from the 62.5 mm (2.5 in.) hydrant outlet using the approach by Walski (1995). The results are shown graphically in Figures 5 and 6, for the strong and weak system, respectively. The results showed that when the system was strong and there was very little head loss other than at the tap and the hydrant, the reduction in hydrant discharge happened at lower values of. Therefore, the impact of a tap that is smaller than the full line size was more significant in the stronger system. The results showed that the percent reduction in flow was roughly independent of head available. Because the percent reduction in flow was independent of available head, it was possible to plot the reduction in flow of a tap compared with a T-fitting over a range of ABOUT THE AUTHORS: Thomas M. Walski* is vice-president of engineering with Haestad Methods Inc., 37 Brookside Road, Waterbury, CT 06708; e-mail <twal@haestad.com>. He has a BA degree from King s College in Wilkes-Barre, Pa., and MS and PhD degrees from Vanderbilt University, Nashville, Tenn. He is a member of AWWA, ASCE, and WEF. He has recently published a book, Water Distribution Modeling. He also has been published in JOURNAL AWWA, Journal WRPM, Journal of Environmental Engineering, Journal of Hydraulic Engineering, Water Resources Bulletin, and Water Environment Research. Brian Lubenow is a graduate student at the University of Delaware, and Jeffrey Spaide is employed by Carroll Engineering. If you have a comment about this article, please contact us at <journal@awwa.org>. REFERENCES ASCE Task Force on Flow in Large Conduits, 1965. Factors Influencing Flow in Large Conduits. Jour. Hydraulic Div., 81-HY11. Crane Company, 1969. Flow of Fluids Through Valves, Fittings, and Pipe. Crane Company, New York. Hydraulic Institute, 1979. Engineering Data Book, Cleveland. Idelchik, I.E., 1999 (3rd ed.). Handbook of Hydraulic Resistance. Begell House, New York. Karney, B.W., 2000. Hydraulics of Pressurized Flow. Water Distribution Systems Handbook (L.W. Mays, editor). McGraw-Hill, New York. Miller, D.S., 1978, Internal Flow Systems. BHRA Fluid Engineering Series, United Kingdom. Roberson, J.A. & Crowe, C.T., 1997. Engineering Fluid Mechanics. John Wiley & Sons, New York. Villemonte, J.R., 1977. Some Basic Concepts on Flow in Branching Conduits. Jour. Hydraulic Div. ASCE, 103-HY7. Walski, T.M., 1995. An Approach for Handling Sprinklers, Hydrants, and Orifices in Water Distribution System Models. 1995 AWWA Ann. Conf., Anaheim, Calif. Walski, T.M, 1984. Analysis of Water Distribution Systems. Van Nostrand Reinhold, New York. 96 JANUARY 2002 JOURNAL AWWA 94:1 PEER-REVIEWED WALSKI ET AL