NEAR REAL-TIM E WATER RESOURCE M ODELLING David Fuller, Principal - - PowerPoint PPT Presentation

NEAR REAL-TIM E WATER RESOURCE M ODELLING

David Fuller, Principal Water M anagement & Technology Angus Swindon, National Director Power and Water

> power generation > flood forecasts > irrigation & power > water management

• Challenges and Future Directions
• Entura >10 years experience in near real-time modelling
• Drivers - M ature system
• Optimisation (water, energy, CO2)
• Operational decisions
• Dam safety
• Lessons learned illustrated by four examples:

 Hydro-Tasmania Operations  Rowallan Dam Upgrade  M eander Dam Optimisation  M odular Software Systems

OVERVIEW

HYDRO-TASM AN IA SYSTEM S

• 30 power stations
• 45 major lakes
• 13,500 M m3/ yr

Inflow Forecasts:

• Dynamic models

(Sharma et al 2005)

• 7 day forecasts
• 61 locations
• 2 hourly updates
• Input: Rainfall
• Input: Streamflow
• Input: SCADA
• Weather forecasts

W HY, W HAT & RESU LTS?

• Increased value of data and information
• Needs

 Reliable data capture  Rapid transfer from field  Automated quality checking  Integration with modelling systems

• Led to:

 Electronic data capture  M ultiple communication systems  Remote monitoring and alarms  Ajenti Data M anagement System (ADM S)  Robust data storage  Improved modelling with dynamic data interfaces

EXAM PLE 1: HYDRO TAS FORECAST U PGRADES

• Small, steep, wet & responsive catchments
• Key limitations of past weather forecasts

 Over-reliance on forecasters  Semi-quantitative, district scale  Limited coverage (only some HT catchments)

• Emergence of Australian Digital Forecast Database :

 Gridded data  Two forecasts per day  Statewide coverage  Biased – downscaling & observed to forecast  Still influenced by forecaster “experience”

RAIN FALL FORECAST BIAS CORRECTION

Daily 25% POE Daily 50% POE 3 hour M ean

100 100 100 100 100

Observed

Day 0 Day 1 Day 2 Day 3 Day 4

Forecast

RESU LTS / CON CLUSION S

• Gridded rainfall data is emerging technology.
• Significant improvement on previous forecasts.
• Good reliability and consistency of forecasts.
• Needs careful bias correction.
• Not a substitute for local observations.
• Careful adoption and adaptation of data is appropriate.
• Recent studies suggest forecasters may be over-riding

large modelled rainfall forecasts!

• What opportunities to remove the forecaster for
• perational model purposes?

EXAM PLE 2: ROW ALLAN REFURBISHM EN T

LAKE ROW ALLAN U PGRADE

Need:

• 43m clay-core dam
• High hazard
• Piping failure at interface

right spillway wall & embankment

• Replace 2 sections of dam

– core & filters.

• Reconstruct upper portion
• f dam crest
• M inimise risks during

construction – safety & cost. Solution:

• Sheetpile wall
• Lake drawdown
• Emergency backfill
• Customised flood warning

system + trigger or review actions.

• M odel performance critical:

+ reliability/ redundancy. + accuracy. + avoid false alarms + avoid costs.

ROW ALLAN – CU STOM FLOOD FORECASTS

• Development of bespoke flood warning model using

Hydstra M odelling.

• Rainfall network and sensitivity testing, alarms and

replacement as necessary (Ajenti).

• M odel constructed based on back calculation of

recorded storage levels and discharges plus rainfall records (20 years).

• Shuffled complex evolution algorithm for parameter
• ptimisation.
• Commercial contract with BoM for numerical weather

forecasts and briefings on weather conditions.

• Customised bias correction.

ROW ALLAN – M ODEL OU TCOM ES

• 10 day probabilistic forecasts with variable trigger levels

depending on stage of construction.

• Tolerances within –0.22m and +0.28m storage level.
• Disseminated by SCADA and SM S to operators, dam

safety engineers and site personnel.

• End-to-end test of model prior to start of work.
• Excellent performance.
• Rain gauge replacement triggered.

EXAM PLE 3: M EAN DER DAM

Need:

• Reliable irrigation delivery
• M inimise releases
• Demonstrate efficiency
• M aximise hydropower
• M onitor water quality
• M aintain environment
• Report to regulators, etc.
• Invoicing
• Flexible / adaptable solution

Solution:

• Ajenti system
• Links to SCADA, BoM ,

DPIWE, HT gauges & water meters

• Optimise within constraints
• OPSIM model
• Customised interface(s)
• Transparent operation
• IE Aust National Award

M EAN DER DAM – IRRIGATION & POW ER

Ajenti TRX Ajenti TRX M eander Dam Irrigation Network M eander Dam SCADA S ystem Private 3G Hosted Network Bureau of M eteorology “ The Cloud” Ajenti ADM S Entura: M anage M onitor Respond Irrigation Operations OPSIM M odel Ajenti Databridge

Alarms SM S, email Forecast Data

EXAM PLE 4: AFRICAN W ATER M AN AGEM EN T

Need:

• Integrated data capture
• M ultiple data sources
• Robust data management
• Integrated with modelling
• Near-term forecasts
• Long-term forecasts
• Invoicing
• Flexible / adaptable solution

Solution:

• Use existing tools
• Careful design stage
• Fit for purpose
• Open, modular approach
• Redundancy built-in
• Customisable interfaces
• Adaptable to latest research
• Consider long term support

African Water M anagement System

Dashboards & Invoicing Drought Forecasts Flood Forecasts Browser Access Data & Graphics M obile Apps Water M anagement Plans

Requirements: Data Sources? Solution?

Satellite Data Products Water M eters M obile Field Data Loggers T elemetry

SCADA

M et. Forecasts Aquarius Time Series Ajenti DM S & Ajenti DataBridge AFDM Aquarius Forecast etc. eWater Source etc. Data Transformation Aquarius WebPortal

W HAT HAVE W E LEARN ED?

• Increasing value of data & information.
• Data collection – from regular to reactive

– from silos to sharing.

• Robust systems increase data access, reliability and use.
• Numerical weather predictions & satellite data

 Increasing availability.  Enhance local data collection.  Not alternatives or replacements.

• M odular software and modelling

 Don’t reinvent the wheel!  Ajenti Data M anagement System (ADM S)

CHALLEN GES AHEAD

• Focus on tools for water managers and operators

 What do they need?  How will they use it?  Don’t forget the need to review history!

• Ensemble forecasts

 data, speed, processing power.

• What role for forecasters?
• M odular software and modelling

 Open systems using appropriate models.  Secure data transfer  Software maintenance and support.

THAN KYOU

David Fuller – Principal Water M anagement & Technology E: david.fuller@ent ura.com .au P: +61 438 559 763 Angus Swindon – Director Power and Water E: angus.sw indon@ent ura.com .au P: +61 6245 4335