H2OMAP Designer Software and the Network Benefits of Integrated Water Management 2013 Asia Pacific Water Industry Modelling Conference
Introduction & Presentation Scope What is H2OMap Designer: H2OMap Designer is a module that comes with H2OMap. It can be used to select optimal pipe sizes. Use of OGA base Software is Not New: However, typically such software is used by consultants and not directly by authorities. MWC Experience: This presentation is based on a MWC case study and our experiences with the application of H2OMap Designer. Case Study of the Process: Focusses on the process of using the H2OMap Designer rather than the actual results of the project. Not an OGA Explanation: Is not an explanation of how H2OMap Designer (and OGAs) works and the algorithms that are used.
Presentation Overview About Melbourne Water (MWC) Project Overview Process Ease of Use Benefit 1 - Unbiased Benefit 2 - Efficiency Benefit 3 - Robust IT Considerations Insights Project Results Conclusions
About Melbourne Water Victorian Government Owned Services: Drinking Water Recycled Water Sewerage Services Drainage Flood Protection Rivers & Creek Protection Assets Managed Include: Water Distribution Mains Dams & Service Reservoirs Streams, Creeks & Rivers Hydro Electric Power Stations Net Gain Sites Jetties Customers: Residents Businesses Bulk Water Customers: City West Water South East Water Yarra Valley Water Western Water Gippsland Water Barwon Water South Gippsland Water Westernport Water Govt. Departments Include: Growth Area Authority Office of Living Victoria
Scope & Purpose of Network Benefits of IWM Project Purpose Comparator Cost Model Customer Enquiries Customer Business Cases Scope Potable Water Costs Only Transfer System Costs Only RWCs know distribution and alternate water costs
Project Approach A High Level Scenario Comparison High Level Scenario Comparison: Scenario A: Base Case (BC) Scenario B: with Integrated Water Management (IWM) Network Benefit IWM = Cost Scenario A Cost Scenario B Scenario Details: 2062 Potable Water Demands Potable Water Demands lower under IWM than BC. Melbourne's Western Growth Corridor Model incidentally includes CBD and part Northern growth Corridor Pipes, Pumps and Tanks considered East West Seasonal Transfer System considered
Approximate Study Area
Model Schematic (Demand Nodes in Yellow & Red) Dynamic simulation required because of inter reservoir transfer. Melton West Werribee Greenvale Reservoir CBD 113 Links 103 Nodes 253 km Pipe DN300 to DN2100 New Pipes Existing Pipes Developed Area Developing Areas MWC Assets CWW Assets WW Assets
Greenvale Boundary Condition Melton Greenvale Reservoir CBD West Werribee
Le Cotchin Boundary Condition Melton Greenvale Reservoir CBD West Werribee
Preston Reservoir Boundary Condition Melton Greenvale Reservoir CBD West Werribee
Illustrative Demand Areas Inner City Development Melton Greenvale Reservoir CBD West Werribee
Illustrative Demand Areas Western Growth Corridor Melton Greenvale Reservoir CBD West Werribee Illustrative Only: This area also contains undevelopable grasslands of high environmental value and areas outside the Urban Growth Boundary.
Indicative Development Forecast for Study Area Year Dwellings 2012 580,000 2062 1,250,000
Model Demands
Actual Sequence of Modelling Process (Not Necessarily Optimal) 1. Discussion with Colleague who had used OGA (~ 1 Hour) 2. Construction of a Simple Trial/Play Model (~ 1 Day) 3. Base Model Selection (~ 1 Hour Discussion) 4. Pipe, Pump & Control Simplification (~5 Days) 5. Troubleshooting Model Runs (~ 3 Days ) 6. Model Refinement following Successful Run (~ 2 Days) 7. Preliminary Simulation Results (Model Runs ~ 1 to 4 Days) 8. Review & Assumption Updates with Customers (~ Weeks) 9. Final Simulation Results (Model Runs ~ 1 to 4 Days)
Process Learning Start with a Discussion & Small Model: A simple play model and a discussion with someone who has used Designer is invaluable. Keep Controls Simple: Simplification is appropriate for a 50 year study. Current operation does not reflect future operation. Simplification, Understanding & Transparency: Simplification improves understanding and makes data/assumptions more robust. Software was very Easy to Use: Don t rule out using the software based on preconceived ideas of complexity. Setting Model Up for OGA Aids Learning: Better appreciation of system constraints. Non-convergence highlighted a missing asset.
Ease of Use Only a few relatively easy to understand screens to be completed: Pipe Groups (Min & Max Diameter, Cost and Pipe IDs) Pipe Actions (New, Parallel or Replace & Unit Costs) Constraints (Junction ID & Min Pressure) Some possible improvements: There was some key punching required to enter the pipe groups into the software (no facility to cut and paste from a spread sheet). Parallel pipes need to be converted to an equivalent diameter in an MS Excel spreadsheet before being added back into the model. Understanding what the OGA parameters mean in real world terms. The H2OMap help manual doesn t really help.
Pipe Groups
Pipe Actions (Costs)
Constraints (Bulk Water Supply Agreement Service Levels)
Model Simulation Screen
Model Output (Exportable to MS Excel)
A perspective on Bias One can only give an unbiased opinion about things that do not interest one, which is no doubt the reason an unbiased opinion is always valueless. The man who sees both sides of a question is a man who sees absolutely nothing. Oscar Wilde
Removal of Bias - Example Variables & Assumptions Variable / Assumption Base Case with IWM Demand Assumptions^ Higher Lower Demand Profiles^ Peakier Flatter Demand Locations Identical Identical Tank Level Control Identical Identical Boundary Conditions Identical Identical Roughness Factors 0.6 mm 0.6 mm Pressure Constraints Explicit & Identical Explicit & Identical Pipe Lengths Identical Identical Pipe Alignments Identical Identical Unit Costs Explicit & Identical Explicit & Identical Pipe Selection Same Process - OGA Same Process - OGA ^ Provided and/or reviewed by retail water customers.
Efficiency Difficult to Assess Time Saving: The alternative of engineering trial and error was not undertaken to compare. More Set Up Time: Using H2OMap Designer arguably requires slightly more set up time than using engineering trial and error. Less Scenario Analysis Time: Once the model is takes less than ½ a day engineering time to change results and get new results. Less Engineering Time Overall: And the ability to easily run additional scenarios in the future allows for better customer service. When to Stop: The OGA based Designer software knows when to stop searching for an optimised answer, some engineers might not.
Robust Results Check Simulation Runs: The same answers were obtained through multiple simulation runs. High Level Review: The final results appear sensible, i.e. there is a certain logic to what the computer has chosen. Pressure Constraints: The pressure constraints were just met which indicates that pipes have generally not been oversized. Ad-Hoc Testing: Limited ad-hoc testing of the Designer solution through inspection (further reducing pipe sizes) was undertaken. Arguably More Transparent: An engineer might find it difficult to articulate what search algorithm was applied/used.
Information Technology Considerations Model Run Time (between 1 4 Days) Use of a Virtual Computer Issues with Computer Energy Savers and Multi-day Simulations Challenges associated with Management of Specialist Software
A way to think about answers produced by H2OMap Designer Even when alternative views are clearly wrong, being exposed to them still expands our creative potential. After hearing someone shout out an errant answer, we work to understand it, which causes us to reassess our initial assumptions and try out new perspectives. Johan Lehrer
Insights arising from H2OMap Designer Decisions 1. Impact of Demand Reduction: Pipes in Inner Melbourne appear to have spare Capacity because of reduction in peak demand. 2. Need for Greenvale Booster WPS: Designer quickly confirmed the need for the recently constructed Greenvale Booster Pump Station. 3. Source Water: H2OMap Designer indicates that demand is supplied via Greenvale and not Western Transfer Main. 4. Dynamic Simulation & Operation: H2OMap Designer managed a dynamic simulation and has potential for control optimisation studies.
Example Base Case Results Red = New Pipe Blue = Pipe Duplication Green = No Change
Example Results with IWM Red = New Pipe Blue = Pipe Duplication Green = No Change
Example Model Results $981M - cost of new potable water pipes under the base case. $730M cost of new potable water pipes with IWM. Approximately a 26% reduction in potable water pipe costs. Important: The cost analysis does not include the cost of the alternate water system that is required to obtain the reduction. These results will from the basis for an input into retail water customer business cases.
Project Benefits of OGA The following benefits were realised: unbiased results modelling efficiency robust results A note on cost savings: There were no cost savings associated with the use of OGA for this specific project. Both the Base Case and the IWM Scenario were optimised using H2OMap Designer. There was no un-optimised solution for comparison.
Conclusion With recent enhancements to modelling software capability there would appear to be the potential for greater use of optimisation software, such as H2OMap Designer, within the Victorian Water industry by water authorities.
Acknowledgements Abigail Farmer, Melbourne Water Bruce Collins, City West Water Muthu Muthukaruppan, City West Water Andrew Chapman, South East Water Rob Belcher, Western Water Alex Walton, Melbourne Water Ann Pugh, Innovyze