Optimatics models best solution
Artificial Intelligence (AI) in the water industry
The 'next wave' of sophistication in water resource handling will see artificial intelligence 'learn' how to best manage water usage, according to optimization firm, Optimatics.
The Fullarton-based firm claims to be leading the world in AI techniques.
"Some have their roots in evolution and survival of the fittest (genetic algorithms)," says Regional Manager Asia Pacific, Dr Paul Smith. "Others are modelled on the way a brain works (neural networks), while yet others replicate the way ants search for food."
Water utilities worldwide are looking to integrate water, wastewater, recycled water and even storm water into networks and improve security of supply, according to Optimatics' Peter Radcliffe.
"How can a water utility choose what to build, where infrastructure should go and how large to build a system?" Peter says.
He says a human mind can come up with a creative solution to a limited engineering problem, but AI can grind away at a large-scale problem 24 hours a day.
In essence, Optimatics software modelling presents variable options which consultants skilled in the field can 'tweak' to obtain best projections for system performance. Many ideas can be tested against the performance objectives, quickly and rigorously.
"The platform also allows all parties, whether they are planners, operators, politicians or finance people to have a say in the constraints and rules that the organisation process is working to satisfy," Peter says.
Genetic algorithms alone in planning a typical water reticulation system can improve hydraulics (service to customers) and reduce cost by 15%.
"Will this spell the end of the engineer? The answer to this question is a firm 'no'," Peter says.
"This is because the pace of change is also increasing and engineers today must manage far more problems in less time than ever before. The AI approach, in fact, offers significant benefits to today's engineers and clients."
In-business magazine, Issue 43, October/November 2008
