Vous êtes sur la page 1sur 20

International Journal of Science and Technology Volume 3 No.

2, February, 2014

IJST 2014 IJST Publications UK. All rights reserved.
101

Statistical Modeling of Head Loss Components in Water Distribution within
Buildings

J. I. Sodiki
Department of Mechanical Engineering
Rivers State University of Science and Technology
P. M. B. 5080
Port Harcourt, Nigeria

ABSTRACT

Fractions of the total head loss which constitute the loss through fittings were computed for varying lengths of first index pipe run,
total flow rate and number of sanitary appliances supplied, in a water distribution system within a building. The results were used in a
regression analysis to obtain second order equations which gave a general increase of the fraction from 0.263 to 0.420 for an increase
in length of first index pipe run of 28.3m to 140.3m; from 0.262 to 0.429 for an increase in the total flow rate of 0.6L/s to 4.4L/s; and
from 0.238 to 0.393 for an increase in the number of supplied sanitary appliances of 8 to 120. The correlation coefficients between the
ratio of loss through fittings to the total head loss and each of the variables of pipe length, flow rate and number of appliances were
0.97, 0.95 and 0.91, respectively; which were found to be acceptable for a 99% confidence level. The ratios are useful for quick estimate
of losses in water distribution systems in buildings, needed in determining the suitability of water reservoir elevations and lift pump
head requirements.
Keywords: Loss ratios, pipe fittings, water distribution within buildings, regression analysis.
1. INTRODUCTION

The available pressure at a given point in a building water
distribution system pipe work is progressively reduced away
from the pressure source (which is usually on elevated
storage). This reduction is due to pipe frictional loss and loss
through pipe fittings such as elbows, tees, reducers and valves.
The latter loss is sometimes called separation loss.

Thus, extensive pipe runs would result in increased frictional
loss, while multiplicity of pipe fittings would be associated
with increased separation loss. However, the number and types
of fittings installed in a given pipe run are specified such as to
achieve system functionality; and it can be assumed that, for a
given system configuration, the ratio between the frictional
loss and the separation loss for an index pipe run may vary with
varying length of pipe run.

In an earlier paper (Sodiki and Orupabo, 2011) it has been
established that the fraction of the total loss which constitutes
the total separation loss increases as the first index pipe length
increases, in simple water distribution systems; an increase in
length of water distribution first index pipe usually implying
increases in the total design water flow rate and in the number
of sanitary appliances being supplied.

In this paper, the ratio of total separation loss to total head loss
is regressed, in turn, on the variables length of first index pipe
run, design flow rate, and number of sanitary appliances, to
obtain equations for predicting the relationship of each variable
with the ratio. Second order relationships are utilized in the
regression analysis, as indicated by the slopes of the curves
obtained earlier (Sodiki and Orupabo, 2011) and the
coefficients of correlation are calculated to check the
usefulness of the resulting equations.
2. HEAD LOSS COMPONENTS
The frictional loss hf and the loss through pipe fittings hp are
analysed as follows:

2.1 Frictional Loss: This loss, for each section of the first
index pipe run, has been derived for plastic pipe
material as (Sodiki, 2002)

hf = 1.1374 x 10
-3
ld
-4.867
q
1.85
- - - - - (1)

where l = pipe length (m)

d = pipe diameter (m)


and q = flow rate (m
3
/s)

Thus, the head loss per metre pipe run is

hf / l = 1.1374 x 10
-3
d
-4.867
q
1.85
- - - - - (2)

A graphical representation of Eqn. 2 (Institute of
Plumbing, 1977) is utilized in the analysis.

2.2 Loss through Fittings: This has been derived for
each fitting as (Sodiki, 2003)

hp = 0.08256kd
-4
q
2

- - - - - (3)

where k is a head loss coefficient of the fitting which is taken
for this analysis as 0.75 for an elbow, 2.0 for a tee and 0.25 for

International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
102

a gate valve (Giles, 1977). For reducers, k is expressed in terms
of the ratio of upstream diameter d1 to downstream diameter d2
as in Table 1 (Giles, 1977); with the k values, so obtained,
being referred to the downstream diameter in Eqn. 3.





Table 1: Values of K for Reducers, in Terms of Ratio of Upstream Diameter (d1) to Downstream Diameter (d2)
(Giles, 1977)


Ratio d1/d2 k
1.2 0.08
1.4 0.17
1.6 0.26
1.8 0.34
2.0 0.37
2.5 0.41
3.0 0.43
4.0 0.45
5.0 0.46
3. THE DISTRIBUTION SYSTEM
CONFIGURATION

The configuration utilized in the analysis is one in which water
is distributed by gravity from a high level reservoir to a range
of identical toilet rooms of a hotel building (Fig. 1). Each room
contains a water closet, a bath tub, a wash basin and a water
heater. The selected pipe work arrangement represents a
commonly occurring scenario in simple water distribution
systems.

3.1 Methods of Estimate
The analysis was first done for the pipe run from A to B and
up to the farthest fixture supplied by the branch from B
(considering the extension on the main distribution pipe from
B towards C as non-existent). In the next step, the analysis was
done for the pipe run from A to C and up to the farthest fixture
supplied by the branch from C (again, considering the
extension from point C towards D as non-existent). Subsequent
steps were carried out in like manner for extended index pipe
runs up to the last step (for the index run from A to P up to the
farthest appliance outlet supplied from P). The progressive
extensions of the first index run provided a variation of the
complexity of pipe work.

The graphical method adopted for pipe sizing and estimation
of frictional losses is illustrated using the pipe run from A to C
and up to pipe section 6 (which was the second step in the
analysis procedure described earlier).

This pipe run is shown as an isometric sketch in Fig. 2 in which
the pipe sections are labeled using boxes. The number to the
left of the box is the pipe section number, the number to the top
right is the measured pipe length (in m), and that on the bottom
right is the flow rate in the section (in L/s).


International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
103




International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
104



International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
105

In the analysis, loading units, which account for the non-
simultaneous use of all the installed sanitary appliances, were
utilized to obtain the flow rates from the graph of loading units
versus flow rates (Fig. 3) (Institute of Plumbing, 1977). These
units were taken as 2 for a water closet cistern, 1.5 for a wash
basin, 10 for a bath tub and 2 for a water heater cylinder.
Cumulative units were, thus, utilized for each pipe section. For
loading units below 10 which are not presented in Fig. 3, linear
extrapolations were made to obtain corresponding flow rates.

Now, for a reservoir height above point A, in Fig. 2, of 10m
and a height of the water heater in pipe section 6 (which is the
final section of this index run) above point A of 2.5m, the
pressure head H available in the first index run = 10m - 2.5m
= 7.5m. The measured length of the index run is L = 36.3m.
Then, the rate of head loss per metre run (H/L) should not
exceed 7.5/36.3 = 0.207m/m run.

This H/L value and the sectional flow rates were used to select
pipe sizes from Fig. 4 (Institute of Plumbing, 1977). For
instance, for pipe section 2 which carries a flow rate of 0.6L/s,
a 25mm pipe was selected (at point A in Fig. 4). The actual
values of H/L were obtained at the intersection of the lines of
flow rate and pipe diameter. For pipe section 2, as an example,
the actual H/L value (at point A in Fig. 4) is 0.085m/m run and
the measured pipe length is 11.0m. Thus, the head loss due to
friction for this pipe section is 0.085 x 11.0m = 0.935m

Table 2 gives a summary of the pipe sizing estimates and the
calculated head losses for this index run.


International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
106


Table 2: Parameters of Distribution System for 36.3m First Index Run, 16 Appliances, 0.95L/s Flow Rate

1 2 3 4 5 6 7 8 9 10 11
Pipe
section
No.
Loading
units
Design
flow (L/s)
Pipe
length (m)
Permissible
maximum
H/L
Dia (mm) Actual H/L Frictional head
loss, hf (m)
Fittings (other than
reducers)
Reducers
(mm x mm)
Loss thru
fittings, hp
(m)
1 62.0 0.95 20.0 0.207 32 0.07 1.400 3 elbows
2 gate valves
1 tee
- 0.338
2 31.0 0.60 11.0 0.207 25 0.085 0.935 3 elbows
2 gate valves
1 tee
32 x 25 0.370
3 19.0 0.45 0.1 0.207 20 0.2 0.02 1 tee 25 x 20 0.220
4 7.0 0.24 2.5 0.207 20 0.065 0.163 1 tee - 0.059
5 3.5 0.12 0.2 0.207 15 0.15 0.03 1 tee 20 x 15 0.050
6 2.0 0.07 2.5 0.207 15 0.06 0.15 1 elbow
1 gate valve
- 0.008
36.3 2.698 1.045

International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
107

With the pipe sizes obtained, locations of reducers in the
first index run were determined. Other types of fitting
(elbows, tees and valves) in the first index run were
specified in consideration of system functionality. In pipe
section 6, for instance, there are one elbow and one gate
valve, such that hp for this section (from Eqn. 3) = (0.75
+ 0.25) x 0.08256 x 0.015
-4
x (0.07 x 10
-3
)
2
= 0.008m,
where d = 0.015m and q = 0.07 x 10
-3
m
3
/s.

Similarly, in pipe section 5, there are one 20mm x 15mm
reducer (which has d1/d2 = 1.33) and one tee k for the
reducer (from Table 1) is 0.139. Then, from Eqn. 3, hp for
pipe section 5= (0.139 + 2) x 0.08256 x 0.015
-4
x
(0.12x10
-3
)
2
= 0.050m







3.2 Resulting Parameters for the Different Index
Runs

In the manner illustrated in section 3.1, the other pipe
configurations which number fourteen were analyzed.
However, in order to maintain brevity, only the isometric
sketches of the shortest and longest first index pipe runs
are presented in Figs. 5a and 5b, respectively. Also,
rather than listing all fourteen tables of calculation
summary corresponding to the fourteen other first index
pipe runs, only the two corresponding to the shortest and
longest runs are presented in Tables 3 and 4, respectively.

Table 5 summarizes the calculated total frictional and
separation losses, as well as the ratios of separation loss
to total loss for the varying complexities of pipe work.
The presentations in Excel graphics of Fig. 6 depict the
variation of the ratio of loss through fittings to total loss
with pipework complexity. Measures of pipework
complexity are presented as length of first index pipe run,
total flow rate from the reservoir and number of sanitary
appliances supplied from the reservoir.


International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
108














International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
109


Table 3: Parameters of Distribution System for 28.3m First Index Run, 8 Appliances, 0.60L/s Flow Rate

1 2 3 4 5 6 7 8 9 10 11
Pipe
section
No.
Loading
units
Design
flow
(L/s)
Pipe
length
(m)
Permissible
maximum
H/L
Dia (mm) Actual H/L Frictional head
loss, hf (m)
Fittings (other
than reducers)
Reducers
(mm x mm)
Loss thru
fittings, hp
(m)
1 31.0 0.60 23.0 0.265 25 0.085 1.955 6 elbows
3 gate valves
1 tee
- 0.552
2 19.0 0.45 0.1 0.265 20 0.200 0.020 1 tee 25 x 20 0.220
3 7.0 0.24 2.5 0.265 20 0.065 0.163 1 tee - 0.059
4 3.5 0.12 0.2 0.265 15 0.150 0.030 1 tee 20 x 15 0.050
5 2.0 0.07 2.5 0.265 15 0.060 0.150 1 elbow, 1
gate valve
- 0.008
28.3 2.318 0.889




International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
110

Table 4: Parameters of Distribution System for 140.3m First Index Run, 120 Appliances, 4.4L/s Flow Rate

1 2 3 4 5 6 7 8 9 10 11
Pipe
section
No.
Loading
units
Design
flow (L/s)
Pipe
length (m)
Permissible
maximum
H/L
Dia (mm) Actual H/L Frictional head
loss, hf (m)
Fittings (other than
reducers)
Reducers
(mm x mm)
Loss thru
fittings, hp
(m)
1. 475.0 4.40 20.0 0.053 65 0.027 0.540 3el, 2g.v 1 tee - 0.426
2. 434.0 4.00 8.0 0.053 65 0.024 0.192 1 tee - 0.148
3. 403.0 3.70 8.0 0.053 65 0.020 0.160 1 tee - 0.127
4. 372.0 3.50 8.0 0.053 65 0.019 0.152 1 tee - 0.113
5. 341.0 3.20 8.0 0.053 50 0.050 0.400 1 tee 65 x 50 0.440
6. 310.0 2.95 8.0 0.053 50 0.045 0.360 1 tee - 0.230
7. 279.0 2.90 8.0 0.053 50 0.043 0.344 1 tee - 0.222
8. 248.0 2.70 8.0 0.053 50 0.040 0.320 1 tee - 0.193
9. 217.0 2.60 8.0 0.053 50 0.035 0.280 1 tee - 0.179
10. 186.0 2.20 8.0 0.053 50 0.026 0.208 1 tee - 0.163
11. 155.0 1.80 8.0 0.053 50 0.020 0.160 1 tee - 0.086
12. 124.0 1.55 8.0 0.053 50 0.016 0.128 1 tee - 0.059
13. 93.0 1.25 8.0 0.053 40 0.045 0.360 1 tee 50 x 40 0.152
14. 62.0 0.95 8.0 0.053 40 0.029 0.232 1 tee - 0.058
15. 31.0 0.60 11.0 0.053 32 0.030 0.330 3el, 2g.v 1 tee 40 x 32 0.138
16. 19.0 0.45 0.1 0.053 25 0.050 0.005 1 tee 32 x 25 0.091
17. 7.0 0.24 2.5 0.053 25 0.017 0.043 1 tee - 0.024
18. 3.5 0.12 0.2 0.053 20 0.020 0.004 1 tee 25 x 20 0.022
19. 2.0 0.07 2.5 0.053 15 0.037 0.093 1 el, 1g.v 1 tee 20 x 15 0.009
140.3 4.311 2.880


International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
111


Table 5: Ratios of Loss through Fittings to Total Loss for Varying Pipe Work Complexities

1 2 3 4 5 6 7
Length of 1
st
index
pipe run (m)
Total flow rate through
main distribution pipe
(L/s)
No. of appliances served by
main distribution pipe
Frictional loss in 1
st
index
run (m)
Loss through fittings in 1
st

index run (m)
Total loss in 1
st
index
run (m)
Ratio of loss through
fittings to total loss
28.3 0.60 8 2.318 0.889 3.207 0.277
36.3 0.95 16 2.698 1.045 3.743 0.279
44.3 1.25 24 3.943 1.302 5.245 0.248
52.3 1.55 32 3.747 1.302 5.049 0.258
60.3 1.80 40 4.777 1.594 6.371 0.250
68.3 2.20 48 4.337 1.627 5.964 0.273
76.3 2.60 56 4.245 1.724 5.969 0.289
84.3 2.70 64 4.625 1.936 6.561 0.295
92.3 2.90 72 5.005 2.069 7.074 0.282
100.3 2.95 80 4.304 2.079 6.383 0.326
108.3 3.20 88 4.379 2.318 6.697 0.346
116.3 3.50 96 4.467 2.666 7.133 0.374
124.3 3.70 104 4.147 2.542 6.689 0.380
132.3 4.00 112 4.059 2.658 6.717 0.396
140.3 4.40 120 4.311 2.880 7.191 0.401



International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
112

Loss Through Fittings
Total Loss
0
0.1
0.2
0.3
0.4
0.5
0 10 20 30 40 50 60 70 80 90 100 110 120 130
No. of Appliances (X
3
)
Loss Through Fittings
Total Loss
0
0.1
0.2
0.3
0.4
0.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Design Fllow Rate (X
2
), L/s
Loss Through Fittings
Total loss
0
0.1
0.2
0.3
0.4
0.5
0 20 40 60 80 100 120 140 160
Index Pipe length (X
1
), m

















Fig. 6 : Variation of Ratio of Loss Thru Fittings to Total Head Loss with Pipework Complexity

International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
113

4. STATISTICAL ANALYSIS OF ESTIMATES
The statistical tool of regression analysis was employed to
obtain equations which represent the variation of the ratio of
fitting loss to total loss with the different measures of pipe
work complexity. Thus, this ratio, denoted as y, was regressed
on the mentioned three measures of complexity, denoted as x1,
x2 and x3, respectively.
An initial investigation had indicated a second order variation
of y with x (Sodiki and Orupabo, 2011) namely
y = a0 + a1 x + a2 x
2

- - - - - (4)
The effort here, therefore, would be to obtain the regression
parameters a0, a1 and a2 for each of the independent variables
x1, x2 and x3 by the solution of the set of analytical
simultaneous equations (Lipson and Seth, 1973)
y = na0 + a1 x+ a2 x
2

yx = a0 x + a1 x
2
+ a2 x
3

- - - - - (5)
yx
2
= a0 x
2
+ a1 x
3
+ a2 x
4


where n = number of data points (in this case, equal to
15).In line with standard methods of statistics, Table 6 was set
up to summarize the computations which aid the analysis. In
Appendix 1, the relationship between y and x1, y and x2, and y
and x3 are established, respectively, as

y =0.2280 1.246 x 10
-3
x1 + 1.558 x 10
-5
x1
2 - --
(6)
y =0.2816 3.538 x 10
-2
x2 + 1.565 x 10
-2
x2
2
--- (7)
and
y =0.2287 + 1.169 x 10
-3
x3 + 1.649 x 10
-6
x3
2
--- (8)


International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
114

Table 6: Compilation of Statistical Variables and Terms

Index
Pipe
Length
X1 (m)
Design
Flow
X2 (l/s)
No. of
Appliances
X3
Frictiona
l Loss
(m)

Loss
thru
Fittings
(m)
Total
Loss (m)
Loss Thru
Fittings/ Total
Loss Y
yx1 x1
2
yx1
2
X1
3
x1
4
yx2 x2
2
yx2
2
x2
3
x2
4
yx3 x3
2
yx3
2
x3
3
x3
4

28.3 0.60 8 2.318 0.889 3.207 0.277 7.8391 800.89 221.842 22665.187 641424.792 0.1662 0.3600 0.09772 0.2160 0.1296 2.216 64 17.728 512 4096
36.3 0.95 16 2.698 1.045 3.743 0.279 10.1277 1317.69 367.636 47832.147 1736306.936 0.2651 0.9025 0.25180 0.8574 0.8450 4.464 256 71.424 5832 65536
44.3 1.25 24 3.943 1.302 5.245 0.248 10.9864 1962.49 486. 698 86938.307 3851367.000 0.3100 1.3625 0.38750 1.7531 2.4414 5.952 576 142.878 13824 331776
52.3 1.55 32 3.747 1.302 5.049 0.258 13.4934 2735.29 705.705 143055.667 7481811.384 0.3999 2.4025 0.61985 3.7239 5.7720 8.256 1024 264.192 32768 1048576
60.3 1.80 40 4.777 1.594 6.371 0.250 15.0750 3636.09 909.023 219256.227 13221150.49 0.4500 3.2400 0.81000 5.8320 10.4976 10.000 1600 400.000 64000 2560000
68.3 2.20 48 4.337 1.627 5.964 0.273 18.6459 4664.89 1273.515 318611.987 21761198.71 0.6006 4.8400 1.32132 10.6480 23.9256 13.104 2304 628.992 110592 5308416
76.3 2.60 56 4.245 1.724 5.969 0.289 22.0507 5821.69 1682.468 444194.947 33892074.46 0.7514 6.7600 1.95364 17.5760 45.6976 16.184 3136 906.304 175616 9834496
84.3 2.70 64 4.625 1.936 6.561 0.295 24.8685 7106.49 2096.415 599077.107 50502200.2 0.7965 7.2900 2.15055 19.6830 53.1441 18.880 4096 1208.320 262144 16777216
92.3 2.90 72 5.005 2.069 7.074 0.292 26.9516 8519.29 2487.633 786330.467 72578302.1 0.8468 8.4100 2.45572 24.3890 70.7281 21.024 5184 1513.728 373248 26873856
100.3 2.95 80 4.304 2.079 6.383 0.326 32.6978 10060.09 3279.589 1009027.027 101205410.8 0.9617 8.7025 2.83702 25.6724 75.7335 26.080 6400 2086.400 512000 40960000
108.3 3.20 88 4.379 2.318 6.697 0.346 37.4718 11728.89 4050.196 127023.787 137566860.6 1.1072 10.2400 3.54304 32.7680 104.8576 30.448 7744 2679.424 681772 5996536
116.3 3.50 96 4.467 2.666 7.133 0.374 43.4962 13525.69 5058.608 1573037.747 182944290.0 1.3090 12.2500 4.58150 42.8750 150.0625 35.904 9216 3446.784 884736 84934656
124.3 3.70 104 4.147 2.542 6.689 0.380 47.2340 15450.49 5871.186 1920495.907 238717641.2 1.4060 13.6900 5.20220 50.6530 187.4161 39.520 10816 4110.080 1124864 116985856
132.3 4.00 112 4.059 2.658 6.717 0.396 52.3903 17503.29 6931.303 2315685.267 306365160.8 1.5840 16.0000 6.33600 64.0080 256.0000 44.352 12544 4967.424 1404928 157351936
140.3 4.40 120 4.311 2.880 7.191 0.401 56.2603 19684.09 7893.320 2761677.827 387463399.1 1.7644 19.3600 7.76636 85.1840 374.8096 48.120 14400 5774.400 1728000 207360000
=
1264.5
=
38.3
= 960 = 4.684 =
419.5892
=
124517.35
=
43315.142
=
13518124.61
=
1559928599.5
=
12.7188
= 116.01 =
40.31322
=
286.0308
=
1361.5298
=
324.504
=
79360
=
28218.048
=
7374536
=
730365952



International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
115

Furthermore, the coefficient of correlation is given as (Lipson
and Seth, 1973)
r =
2
.
1
|
|
.
|

\
|

y
x y
S
S

(9)


where Sy.x =
( )
3
2
1

=
n
y y
n
i
ic i

(10)
= standard error of estimate,
yi being the actual values
of y, yic being the values of
y computed from the
regression equation and n
the numbers of points
and Sy =
( )
1
2
1

=
n
y y
n
i
i


(11)
= sample standard deviation
of y
n 3 is the number of degrees of freedom, as the number of
regression parameters to be estimated is three, a0, a1 and a2
In Appendix II, the coefficients of correlation between y and
x1, y and x2, and y and x3 are computed, respectively, as 0.97,
0.95 and 0.91.
5. DISCUSSION OF RESULTS
From statistical tables (Lipson and Seth, 1973), r required for
99% confidence is 0.661. Since all three computed values of r
exceed 0.661, there is 99% confidence that the variation of the
fraction of head loss through fittings is interdependent, in turn,
with variations of length of index pipe run, total flow rate and
number of appliances served. It further follows that estimates
of the fraction of head loss can be made using the derived
regression equations.
The fractions of head loss through pipe fittings calculated by
the regression equations in Appendix II show the following
general increases: from 0.263 to 0.420 for increase in first
index pipe length from 28.3m to 140.3m; from 0.262 to 0.429
for an increase in total flow rate from 0.6L/s to 4.4L/s; and
from 0.238 to 0.393 for an increase in number of sanitary
appliances from 8 to 120.
It is, thus, observed that the head loss fractions fall between
0.24 and 0.43 within the limits of system complexity utilized
in the analysis.
In fact, random computations and estimates done on other
water distribution configurations (different from the general
architectural arrangement of toilet rooms and sanitary
appliances employed in this analysis), but utilizing the same
available head of 7.5m and distributing to a single building
floor, as in this analysis, had given fractions of head loss in the
region of 0.2 to 0.4 for first index lengths of between 20m and
120m. Hence, needed estimates of fractions may be made by
interpolating between these limits and adding a safety margin.
Alternatively, the present results may be applied with
reasonable correctness.
6. CONCLUSIONS
Within the limits of system complexity of up to 140m index
pipe length, 4.4L/s flow rate and 120 sanitary appliances a ratio
of loss through fittings to total loss of 0.45 is not likely to be
exceeded for simple water distribution systems under an
available head of 7.5m.
The ratios of head loss obtained from the regression equations
are useful in the analysis of system losses. For instance, for a

International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
116

given first index pipe length, the total head loss can quickly be
estimated by adding the relevant fraction to the frictional loss.
The total head loss, so estimated, is useful in determining the
suitability of the given reservoir elevation and, hence, in
assessing the head requirements of the lift pump.
Other water distribution systems such as are utilized for groups
of buildings in housing estates and for distribution in town and
village mains could also be analyzed by the same procedure.
Furthermore, the relationship between the available head and
the computed ratios could be investigated by carrying out the
foregoing analysis for different commonly utilised reservoir
elevations
REFERENCES
[1]. Barry, R. (1977). The Construction of Buildings, Vol. 5:
Supply and Discharge Services. Granada Publishing Ltd,
London.
[2]. Giles, R. V. (1977), Fluid Mechanics and Hydraulics.
McGraw-Hill Book Co., New York
[3]. Institute of Plumbing (1977). Plumbing Services Design
Guide, Essex
[4]. Lipson, C and Seth, N. J. (1973). Statistical Design and
Analysis of Engineering Experiments. McGraw-Hill
Book Co., New York
[5]. Sodiki, J. I. (2002). A Representative Expression for
Swimming Pool Circulator Pump Selection. Nigerian
Journal of Engineering Research and Development. Vol.
1, No. 4, Pp. 24-35
[6]. Sodiki, J. I. (2003). Design Analysis of Water Supply
and Distribution to a Multi-Storey Building Utilizing a
Borehole Source. Nigerian Journal of Industrial and
Systems Studies. Vol. 2, No. 2, Pp 16-32
[7]. Sodiki, J. I. and Orupabo, S. (2011). Estimating Head
and Frictional Losses through Pipe Fittings in Building
Water Distribution Systems. Journal of Applied Science
and Technology. Vol. 16, Nos. 1 & 2, Pp 67-74









International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
117

APPENDIX I: REGRESSION EQUATIONS
(i) Regression of y on x1
Substitution of values from Table 6 into Eqn. 5 yields the simultaneous equation
4.684 = 15a0 +1264.5a1 + 124517.35a2 ----- (a)
419.589 = 1264.5a0 + 124517.35a1 + 13518124.61a2 ---- (b)
43315.142 = 124517.35a0 + 13518124.61a1 +1559928599.5a2 --- (c)
Solving for a0, a1 and a2 yields the expression for Eqn. 4 as
y = 0.2280 1.246 x 10
-3
x1 + 1.558 x 10
-5
x1
2

(ii) Regression of y on x2
Substitution of values from Table 6 into Eqn. 5 yields the simultaneous equations
4.684 = 15a0 +38.3a1 + 116.01a2 ----- (d)
12.719 = 38.3a0 + 116.01a1 + 1386.03a2 ---- (e)
40.313 = 116.01a0 + 386.03a1 +1361.53a2 ---- (f)
Solving for a0, a1 and a2 yields the expression for Eqn. 4 as
y = 0.2816 3.538 x 10
-2
x2 + 1.565 x 10
-2
x2
2

(iii) Regression of y on x3
Substitution of values from Table 6 into Eqn. 5 yield the simultaneous equations
4.684 = 15a0 +960a1 + 79360a2 ----- (g)
324.508 = 960a0 + 79360a1 + 7374536a2 ---- (h)
28218.048 = 79360a0 + 7374536a1 +730365952a2 ---- (i)
Solving for a0, a1 and a2 yields the expression for Eqn. 4 as
y = 0.2287 + 1.169 x 10
-3
x3 + 1.649 x 10
-6
x3
2

APPENDIX II: CALCULATION OF COEFFICIENTS
II(i) Coefficient of Correlation for x1
The regression equation is
yc = 0.2880 1.246 x 10
-3
x1 + 1.558 x 10
-5
x1
2


International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
118

where, yc is the calculated value of y. In Table A1, A2 and A3 yi is the actual value of y obtained from the analysis of the losses of
pressure head, and y is the mean of the y values.
Table A1: Values for Calculating Correlation Coefficient for x1
i xi yi
yi y (yi y )
2
x 10
4

yic yi yc (yi yic)
2

x 10
4

1 28.3 0.277 -0.0236 5.5696 0.265 0.012 1.44
2 36.3 0.279 -0.0333 11.0889 0.263 0.016 2.56
3 44.3 0.248 -0.0526 27.6676 0.263 -0.015 2.25
4 52.3 0.258 -0.0426 18.1476 0.265 -0.007 0.49
5 60.3 0.250 -0.0506 25.6036 0.270 -0.020 4.00
6 68.3 0.273 -0.0276 7.6176 0.276 -0.003 0.09
7 76.3 0.289 -0.0116 1.3456 0.284 0.005 0.25
8 84.3 0.295 -0.0056 0.3136 0.294 0.001 0.01
9 92.3 0.292 -0.0086 0.7396 0.306 -0.014 1.96
10 100.3 0.326 0.0254 6.4516 0.320 0.006 0.36
11 108.3 0.346 0.0454 20.6116 0.336 0.010 1.00
12 116.3 0.374 0.0734 53.8756 0.354 0.020 4.00
13 124.3 0.380 0.0794 63.0436 0.374 0.006 0.36
14 132.3 0.396 0.0954 91.0116 0.396 0.000 0.00
15 140.3 0.401 0.1004 100.8016 0.420 -0.019 3.60
= 1264.5 = 433.8893 = 22.38
y = 0.3123

Sy.x = 01366 . 0
12
10 38 . 22
4
=


Sy = 05567 . 0
14
10 8893 . 433
4
=


Coefficient of Correlation r = 97 . 0
05567 . 0
01366 . 0
1
2
=
|
.
|

\
|

II(ii): Coefficient of Correlation for x2
The regression equation is
yc = 0.2816 3.538x 10
-2
x2 + 1.565x10
-2
x2
2



International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
119

Table A2: Values for Calculating Correlation Coefficient for x2
i xi yi
yi y (yi y )
2
x 10
4

yic yi yc (yi yic)
2
x
10
4

1 060 0.277 -0.0236 5.5696 0.266 0.011 1.21
2 0.95 0.279 -0.0333 11.0889 0.262 0.017 2.89
3 1.25 0.248 -0.0526 27.6676 0.262 -0.014 1.96
4 1.55 0.258 -0.0426 18.1476 0.264 -0.006 0.36
5 1.80 0.250 -0.0506 25.6036 0.269 -0.019 3.61
6 2.20 0.273 -0.0276 7.6176 0.280 -0.007 0.49
7 2.60 0.289 -0.0116 1.3456 0.295 -0.006 0.36
8 2.70 0.295 -0.0056 0.3136 0.300 -0.005 0.25
9 2.90 0.292 -0.0086 0.7396 0.311 -0.019 3.61
10 2.95 0.326 0.0254 6.4516 0.313 0.013 1.69
11 3.20 0.346 0.0454 20.6116 0.329 0.017 2.89
12 3.50 0.374 0.0734 53.8756 0.349 0.025 6.25
13 3.70 0.380 0.0794 63.0436 0.365 0.015 2.25
14 4.0 0.396 0.0954 91.0116 0.390 0.006 0.36
15 4.4 0.401 0.1004 100.8016 0.429 -0.028 7.84
= 36.02
Sy.x = 01733 . 0
12
10 02 . 36
4
=



Sy = 0.05567 (as for x1)
Coefficient of Correlation r = 95 . 0
05567 . 0
01733 . 0
1
2
=
|
.
|

\
|

II(iii): Coefficient of Correlation for x3
The regression equation is
yc = 0.2287 + 1.169 x 10
-3
x3 + 1.649 x 10
-6
x 3
2

Table A3: Values for Calculating Correlation Coefficient for x3
i xi yi
yi y (yi y )
2
x 10
4

yic yi yc (yi yic)
2
x
10
4

1 8 0.277 -0.0236 5.5696 0.238 0.039 15.21
2 16 0.279 -0.0333 11.0889 0.248 0.029 8.41
3 24 0.248 -0.0526 27.6676 0.258 -0.070 1.00
4 32 0.258 -0.0426 18.1476 0.268 -0.010 1.00

International Journal of Science and Technology (IJST) Volume 3 No. 2, February, 2014


IJST 2014 IJST Publications UK. All rights reserved.
120

5 40 0.250 -0.0506 25.6036 0.278 -0.028 7.84
6 48 0.73 -0.0276 7.6176 0.289 -0.016 2.56
7 56 0.289 -0.0116 1.3456 0.299 -0.010 1.00
8 64 0.295 -0.0056 0.3136 0.310 -0.015 2.25
9 72 0.292 -0.0086 0.7396 0.321 -0.029 8.41
10 80 0.326 0.0254 6.4516 0.333 -0.007 0.49
11 88 0.346 0.0454 20.6116 0.344 0.002 0.04
12 96 0.374 0.0734 53.8756 0.356 0.018 3.24
13 104 0.380 0.0794 63.0436 0.368 0.012 1.44
14 112 0.396 0.0954 91.0116 0.380 0.016 2.56
15 120 0.401 0.1004 100.8016 0.393 0.008 6.40
= 61.85
Syx = 0227 . 0
12
10 85 . 61
4
=


Sy = 0.05567 (as for x1)
Coefficient of Correlation r = 91 . 0
05567 . 0
0227 . 0
1
2
=
|
.
|

\
|

Vous aimerez peut-être aussi