Design

 of  water  distribution  systems  

The   network   of   pipes   which   makes   up   the   distribution   system   may   be   subdivided   into  
primary  or  arterial  lines,  secondary  lines  and  small  distribution  mains.  
The  primary  or  arterial  mains  form  the  basic  structure  of  the  system  and  carry  flow  from  the  
pumping  station  to  and  from  elevated  storage  tanks  and  to  the  various  districts  of  the  city.    
The  secondary  lines  form  smaller  loops  within  the  primary  mains  and  run  from  one  primary  
line  to  another.    
The   small   distribution   mains   form   a   grid   over   the   entire   service   area   –   supplying   water   to  
every   user   and   to   the   fire   hydrants.   They   are   connected   to   primary,   secondary,   or   other  
small   mains   at   both   ends   and   are   valved   so   that   the   system   can   be   shut   down   for   repairs  
without  depriving  a  large  area  of  water.  The  size  of  the  small  mains  is  generally  dictated  by  
fire  flow  except  in  residential  areas  with  very  large  lots.  
 
The  detailed  design  of  a  water  distribution  system  is  affected  by  local  topography,  existing  
and   expected   population   densities,   and   commercial   and   industrial   demand.   First,   the   flow  
must  be  disaggregated  to  individual  subareas  of  the  system  .  Next,  a  system  of  interlocking  
loops  must  be  laid  out.  The  disaggregated  flows  are  then  assigned  to  the  various  nodes  of  
the  system.  The  design  then  involves  determination  of  the  sizes  of  the  arterials,  secondary  
lines,   and   small   distribution   mains   required   to   ensure   that   the   pressures   and   velocities  
desired  in  the  system  are  maintained  under  a  variety  of  design  flow  conditions.  These  design  
considerations   are   based   on   the   maximum   daily   flow   rate   plus   one   or   more   fires,  depending  
on  the  size  of  the  community.  The  fire  flow  rate  depends  on  the  character  of  the  individual  
subarea.   In   general,   those   fire   location   s   which   are   most   distant   either   vertically   or  
horizontally   from   the   pumping   plant   will   be   critical   for   design;   however   i   tis   usually  
necessary  to  assume  various  fire  locations  in  order  to  ensure  that  all  areas  are  adequately  
protected.  
Consideration   of   the   design   problem   described   above   leads   us   to   the   obvious   conclusion  
that,  in  general,  there  are  many  possible  solutions  which  will  satisfy  the  design  constraints.  
The   task   then   becomes   determining   the   “best”   solution.   Such   an   optimization   problem   for   a  
looped   pipe   network   is   very   complicated   since   the   distribution   of   flows   in   the   pipes   is   a  
function  of  the  design,  hence  simplified  techniques  are  often  used.  Even  so,  the  optima  for  
the  various  design  conditions  will  not  be  the  same,  so  that  the  final  design  which  satisfies  all  
required  conditions  may  not  be  an  optimum  for  any  particular  condition.  
The   usual   engineering   approach   to   design   of   looped   pipe   systems   involves   layout   of   the  
network,   assignment   of   estimated   pipe   sizes   and   calculation   of   resulting   flows   and   head  
losses.   The   pipe   sizes   are   then   adjusted   as   necessary   to   ensure   that   the   pressures   at   the  
various   nodes   and   the   velocities   in   the   various   pipes   meet   the   criteria   established   for   the  
community.  For  a  given  set  of  pipe  sizes,  the  calculation  of  flows  and  pressures  is  normally  a  
reasonably  straightforward  task  which  can  be  performed  in  a  variety  of  ways.  
 
The   Hardy-­‐Cross   method   and   its   modifications   have   been   used   in   design   and   analysis   of  
water  distribution  systems  for  many  years.  The  method  is  based  upon  the  hydraulic  formulas  
previously   studied,   which   are   used   to   calculate   the   energy   losses   in   the   elements   of   the  
system.   It   is   not   unusual   to   neglect   the   losses   in   fittings,   since   these   will   be   small   with  
respect   to   those   in   long   pipes.   The   energy   loss   in   any   element   of   the   system   may   be  
expressed  as  
 
ℎ! = 𝑘! 𝑄!!             (1)  
 
where     hi  =  energy  loss  in  element  i  
  Qi  =  flow  in  that  element  
   ki  =  constant  depending  on  pipe  diameter,  length,  type  and  condition  
  x  =  1.85  to  2  normally,  depending  on  equation  (units)  used  
  Note:  x  =  1.85  for  SI  units  
 
For  any  pipe  in  a  loop  of  the  system,  the  actual  flow  will  differ  from  an  assumed  flow  by  an  
amount  Δ:  
 
𝑄! = 𝑄!! + ∆             (2)  
 
where   Qi  =  actual  flow  in  pipe  
  Qi0  =  assumed  flow  
  Δ  =  required  correction  
 
Substituting  equation  (2)  in  (1)  gives  
 
(!!!)
𝑘! 𝑄!! =   𝑘! 𝑄!!
!
+ 𝑥𝑄!!  ∆ + ⋯       (3)  
 
The  remaining  terms  in  the  expansion  may  be  neglected  if  Δ  is  small  compared  to  Qi.  For  any  
loop,  the  sum  of  the  head  losses  about  the  loop  must  be  equal  to  zero.  Thus,  for  any  loop,  
 
! !
! 𝑘! 𝑄! = 0             (4)  
 
where  n  is  the  number  of  pipes  in  the  loop.  Then,  from  equation  (3)  
 
! ! ! ! ! (!!!)
! 𝑘! 𝑄! =   ! 𝑘! 𝑄!! +   ! 𝑥𝑘! 𝑄!!  ∆  = 0     (5)  
 
Equation  (5)  may  then  be  solved  for  the  correction:  
 
! ! !
! !! !!! ! !!
∆  =   − ! !! !!! =   − !         (6)  
! !! ! ! !! !!!
!
 
The  procedure  (usually  performed  on  a  computer)  may  be  outlined  as  follows:  
1. Disaggregate  the  flow  to  the  various  blocks  or  other  subareas  of  the  community.  
2. Concentrate  the  disaggregated  flows  at  the  nodes  of  the  system.  
3. Add  the  required  fire  flow  at  appropriate  nodes.  
4. Select  initial  pipe  sizes  using  the  defined  criteria  (legal  criteria).  
5. Assume  any  internally  consistent  distribution  of  flow.  The  sum  of  the  flows  entering  
and  leaving  each  node  must  be  equal  to  zero.  
6. Compute   the   head   loss   in   each   element   of   the   system.   Conventionally,   clockwise  
flows  are  positive  and  produce  positive  head  loss.  
7. With  due  attention  to  sign,  compute  the  total  head  loss  around  each  loop:  
 
! !
!
ℎ! =   𝑘! 𝑄!!  
! !
 
8. Compute,  without  regard  to  sign,  the  sum  
 
!
(!!!)
𝑘! 𝑄!!  
!
 
 9. Calculate  the  corrections  for  each  loop  from  equation  (6)  and  apply  the  correction  to  
each  line  in  the  loop.  Lines  common  to  two  loops  receive  two  corrections  with  due  
attention  to  sign.  
10. Repeat  the  procedure  until  the  corrections  calculated  in  step  9  are  less  than  some  
stipulated  maximum.  The  flows  and  pressures  in  the  initial  network  are  then  known.  
11. Compare  the  pressures  and  velocities  in  the  balanced  network  to  the  criteria  (legal  
criteria).   Adjust   the   pipe   sizes   to   reduce   or   increase   velocities   and   pressures   and  
repeat  the  procedure  until  a  satisfactory  solution  is  obtained.  
12. Apply   any   other   fire   flow   conditions   which   may   be   critical   and   reevaluate   the  
velocities  and  pressure  distribution.  Adjust  pipe  sizes  as  necessary.