Genetic Algorithm Optimisation

1
Genetic Algorithm Optimisation
Only the fittest solutions survive By Khaled
Jame Network Development Manager Halcrow Water
Services (HWS)
2
History

• 1990 Prof. Graeme Dandy, Assoc. Prof. Angus
  Simpson and Dr. Laurie Murphy of Adelaide
  University were the first in the world to apply
  this innovative technique to water systems and
  were co-founders of Optimatics Pty Ltd.
• Optimatics commenced operations in 1996 and has
  developed to provide optimisation services in
  Australia and North America.
• Optimatics formed a strategic partnership with
  Halcrow Water Services in 2003
• UK Projects

3
An introduction ..

• The Optimatics Genetic Algorithm (OGA) is a
  mathematical tool which helps engineers make
  complex design and operational DECISIONS. It uses
  a hydraulic model to identify an optimal
  solution.

• HWS Optimatics use the OGA to provide
  consulting services to the Water industry.

4
(There are 710 combinations or 282,475,249
solutions)
5
100x
(5)
(19)
... and so on
6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
GA Optimization process
Step 1 Create a Network Model of the existing
system. Step 2 - Calibrate the model based on
field testing. Step 3 - Establish future water
demands to be met, and design performance
criteria to be satisfied. Step 4 - Identify
allowable choices for new or replacement pipes,
tanks, valves pumps, and operational
choices. Step 5 - Formulate GA routine for those
decision variables. Step 6 - Link the hydraulic
model to the GA routine.
14
GA Optimization process
Step 7 - Perform preliminary and interim GA runs
get feedback and direction from the utility
engineers. Step 8 Re-run OGA to incorporate
changes. Step 9 - Finalize preferred alternative
solutions and verify.
15
Asset Planning Optimisation

• Capital works plans
• Staging of assets
• Select pipes
• Size storages (reservoir)
• Select pumps
• Select valve settings
• Set pressure zone boundaries
• Identify tank volumes
• Size water treatment plants

16
Operations Optimisation

• Reduce power bill
• Avoid peak instantaneous charges
• Improve fire flows
• Review operating procedures
• Reduce pressures
• Flatten out pumping curves during the day

17
Study Cases
1- Norwich Optimisation
18
(No Transcript)
19
Original Model Number of Nodes 12929 Number
Elements 14010
20
Reduced Model Number of nodes 6700 Number of
elements 7217
21
(No Transcript)
22
Formerly open pipes closed Formerly closed pipes
opened Proposed new cross-connections
Heigham WTW
Moushold site
Lakenham Reservoir
23
Other Studies
24
Case Study 2Grand Prairie, Texas
25
Fort Worth
North Dallas
Decommission
Pump Station
Interconnect
1 MG EST
Interconnect

South Dallas

Non-essential pipes
and duplicate pipes
(min. 0" diameter)
Add 17 MG GSTs


New essential pipes
Lake Joe Pool
at site on ridge
(min. 8" diameter)
29 MG GSTs
Rapid growth is
forecast in the
City of Grand Prairie
southern areas
Preferred Year 2015
Optimized Master Plan
2 MG EST
No Midlothian water
26
City of Grand Prairie, Texas

• 117,000 people over 90 square mile service area
• Average day maximum day demands 200 increase
  by year 2015 to 51 MGD and 92 MGD
• 95 of supply from Dallas Fort Worth via three
  connections with remainder from wells
• Aim to identify best Master Plan for 2015
  condition

27
(No Transcript)
28
Results of GA Optimization Run GA6 The sites of
the 2 new elevated tanks selected are shown
(north-west of Traders pressure plane and in the
far south)
Results of GA Optimization Run GA7 The sites of
the 2 new elevated tanks selected by the GA are
shown (both in the developing southern area.
Schematic showing existing tank sites (blue) and
12 possible new tank sites (red)
29
Summary of OGA results for three final solution
alternatives
30
Case Study 3The Palm Jumeirah, UAE

• Man made Island shaped like a Palm
• Located just off the coast of the city of Dubai
  in the UAE
• 2,500 exclusive residential beachside villas,
• 2,400 shoreline apartments,
• 50 Hotels
• restaurants, shopping malls, sports facilities,
  health spas

31
The Palm - Jumeirah
Photo Michelle Sabti
32
The Palm - Jumeirah
Photo Michelle Sabti
33
(No Transcript)
34
Summary

• GA optimisation enables water designs of
  significant complexity to be tackled
• GA optimisation is very flexible and is able to
  deal with whatever the hydraulic simulation model
  is capable of
• A range of different solutions are generated by
  the GA optimisation process
• Significant savings on CAPEX

35

• Thank you
• Any Questions?

36
(No Transcript)