Hanoi, January 29 th 2015 RODOLFO SONCINI-SESSA DEI – Politecnico di Milano IMRR Project 14 – First policy design results INTEGRATED AND SUSTAINABLE WATER MANAGEMENT OF RED-THAI BINH RIVER SYSTEM IN A CHANGING CLIMATE
The policy design GA
The Design Problem (It)(It) scenario (I t, θ) scenario
Design scenario: DS1 20 years of regular flows (1/5/ /4/2010) 19 years - 5 extreme flood years (red) -5 extreme drought years (yellow) -9 normal years interposed (blue) regular indicators extreme indicators J F, J H, J S J eF, J eS
Extreme events
Design scenario Extreme years
Water supply From IMRR wiki website:IMRR wiki website 7 i ws:3 JSJS
Flood control From IMRR wiki website:IMRR wiki website i Fl:1 8 JFJF
Hydropower From IMRR wiki website:IMRR wiki website: i HP:1 JHJH
Pareto front Extreme floods JSJS JHJH JFJF J eF
Pareto front Extreme floods JHJH JFJF J eF
Pareto front Extreme floods JSJS JFJF J eF
Pareto front Extreme floods JSJS JHJH J eF
BAU alternative The historical behavior (releases and levels) are evaluated in the design condition: 1. the river incision is assumed to be the one of 2010 for all the period ; 2. the hydropower demand is the one of 2010; 3. the water supply demand is the one of 2010; 4. Van Coc operating rule is the present one 5. Hoa Binh reservoir only is assumed to be active (Thac Ba data are available from 2002 only!) Then the BAU evaluation does not exactly reflect what happened in the past!
Duong/Hong ratio
Pareto front (Scenario DS1, 2.5 M policy evaluations) * JSJS JHJH JFJF
* JHJH JFJF
JFJF J eF * ?
Alternative Indicator Percentual deficit [%] Percentual deficit of HBinh [%] Hist. HP of Hoa Binh Historical period -> 2 Power plants only!
Pareto front (Scenario DS1, 2.5 M policy evaluations) * JHJH JFJF
Hoa Binh release Turbine capacity
Pareto front (Scenario DS1, 2.5 M policy evaluations) Historical production * * -75 % -83 % JFJF
Pareto front (Scenario DS1, 2.5 M policy evaluations) * -75 % -98 % JSJS JFJF
Pareto front (Scenario DS1, 2.5 M policy evaluations) * -83 % -98 % JHJH * JSJS
Parallel coordinates plot A1156 Hist. Natural
Alternative A design Ha Noi ( ) 11.5 First alarm
Alternative A1156 – design Son La ( )
Alternative A1156 – design Hoa Binh ( )
Alternative A1156 – design Tuyen Quang ( ) The irrigation is supplied by Tuyen Quang reservoir mainly, since the production on Hoa Binh is more valuable: higher head
Design + validation scenario Extreme years Validation
Alternative A validation Son La ( ) 2012
Alternative A validation Hoa Binh ( ) 2012
Alternative A validation Tuyen Quang ( ) 2012
Design + validation scenario Extreme years Validation
Son La inflow
Son La storage
Hoa Binh inflow
Hoa Binh decision
Tuyen Quang inflow
Tuyen Quang storage
Storage space
Trajectories of A1156 in the storage space
Design scenario: DS1 20 years of regular flows (1/5/ /4/2010) 19 years - 5 extreme flood years (red) -5 extreme drought years (yellow) -9 normal years interposed (blue) regular indicators extreme indicators J F, J H, J S J eF, J eS DS2 23 years of regular flows (1/5/ /4/2013)
Pareto front (Scenario DS2, M policy evaluations) JSJS JHJH JFJF
JHJH JFJF
JSJS JFJF
J eF JFJF
What about improper behavior?
Trajectories of B79 in the storage space
Trajectories of A1156 in the storage space
Trajectories of B79 in the storage space
Further 2.5 M policies evaluations (+13 days) were run to ascertain that nothing better can be obtained. In total 28 days of run for a total of years of evaluation
Pareto front (Scenario DS2, policy evaluations) JSJS JHJH JFJF
-75% -99 % * JHJH JFJF
Pareto front (Scenario DS2, policy evaluations) -75 % -98 % * JSJS JFJF
Pareto front (Scenario DS2, policy evaluations) -99 % -98 % * JSJS JHJH
Pareto front (Scenario DS2, policy evaluations) J eF JFJF
Parallel coordinates plot group C C123 A1154
C123: Water level at Ha Noi
Parallel coordinates plot group C C123 A1154 C254
C254: Water level at Ha Noi
C254: Water level at Ha Noi 2000 K scenario
1971 flood and C254
1996 flood and C254
C254: Son La storage design scenario
C254: Son La storage 2000 years scenario
C254: Tuyen Quang storage design scenario
C254: Tuyen Quang storage 2000 years scenario
Validation of the Extreme Splitting Approach J eF JFJF 2000 years scenario 2K scenario
Pareto front group C JSJS JHJH JFJF J eF
Pareto front group C JHJH JFJF J eF
Pareto front group C JSJS J eF JFJF
Pareto front group C JSJS JHJH J eF
Pareto front group C J eF JFJF JHJH
1996 flood and C254
1996 Rivers flows
1996 Son La – Hoa Binh storages
1996 Son La – Hoa Binh decision and release
1996 Son La storage and release
Next week We will analyze in detail the effects of the policies on the different sectors. We will establish the evaluation criteria of each of them.
What next? 1.The information used by the policy is the storages and the time only. What is the advantage of adding other information? E.g. Ha Noi level, the volume stored in the irrigation canal or the current inflows? 2.We have neglected the storage thresholds presently imposed by law. What is the cost and advantage of maintaining them? 3.What is the best form for the policy?
Thanks for your attention XIN CẢM ƠN
Schedules Group A: u First trial with random seeds and termination of 1M evaluations using design scenario of 39 years DS1 (from 1/5/1990 – 30/4/2010 and extreme years ) u Validate some good alternatives of group A, ex. A1156 with three years (1/5/2010 – 30/4/2013) adding to the design scenario and 2000 years scenarios. Group B: u Second trial start from group A and termination after 1.5M evaluations using design scenario of 39 years DS1. u Validate some good alternatives of group B, ex. B094 with three years (1/5/2010 – 30/4/2013) adding to the design scenario and 2000 years scenarios. Need to add also three years (1/5/2010 – 30/4/2013) to design scenario since this period includes three continuos drought year which not happen in the history. The optimal population of Group B is reduced by criteria on five objective scales: J F 15,000 ; J eF 300,000; J H 0.6; J S 150; J eS 450. and have 191 alternatives which will be used as the initial population of group C.
Schedules Group C: u Start from reduced final population of Group B (191 alternatives) and stop after 0.5M evaluations using design scenario of 42 years DS2 (from 1/5/1990 – 30/4/2013 and extreme years) u Validate some good alternatives of group C, ex. C with 2000 years scenarios. Group D: u Continue with the final population of Group C and stop after +1.5 M evaluations using design scenario of 42 years DS2 (from 1/5/1990 – 30/4/2013 and extreme years ) u Validate some good alternatives of group D, with 2000 years scenarios. Group E: u Continue with the final population of Group D and stop after +1M evaluations using design scenario of 42 years DS2 (from 1/5/1990 – 30/4/2013 and extreme years ) u Validate some good alternatives of group E, with 2000 years scenarios. u In total 5.5 M
Threshold violation in the history Illegal violation … … since Ha Noi is below the alarm threshold. Legal gradual level growth
Threshold violation in the design Illegal violations … … since Ha Noi is below the alarm threshold. Legal gradual level growth