slide1
Download
Skip this Video
Download Presentation
From the PIP procedure to MODSSs

Loading in 2 Seconds...

play fullscreen
1 / 32

From the PIP procedure to MODSSs - PowerPoint PPT Presentation


  • 96 Views
  • Uploaded on

MNR L03. From the PIP procedure to MODSSs. Andrea Castelletti. Politecnico di Milano. Planning actions and management actions. Planning actions : decided once forever or over a long time horizon. Management actions : decided frequently or even periodically, often on a daily basis.

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about ' From the PIP procedure to MODSSs' - sirvat


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
slide1

MNR

L03

From the PIP procedure to MODSSs

Andrea Castelletti

Politecnico di Milano

planning actions and management actions
Planning actions and management actions

Planning actions:decided once forever or over a long time horizon.

Management actions:decided frequently or even periodically, often on a daily basis.

How are they taken?

Planning actions:by means of a Project, i.e. by evaluating different alternatives (i.e. mix of planning actions) with the aim of individuating those that better satisfy the DM and/or Stakeholders’ point of views.

Management actions:taken on the basis of the Regulator’s experience, i.e. somehow empirically.

Does not work!!!

planning a new reservoir

inflows

inflows

t

t

capacity

capacity

levels

levels

t

t

releases

releases

t

t

G

D

G

D

Planning a new reservoir

Planning decision:to build the reservoir

Management decision:water volume to be released in the next 24 hours

.......

Deciding to build the reservoir does require deciding how it will be daily regulated, otherwise it is not possible to evaluate if and how the farmers are satisfied.

The management must be always considered when either the planning requires it or it change the context in which the current managemt is performed.

Planning the management

planning the management

Release plan

Planning the management

IDEA: we can define the management decision for each day of the Project horizon (N years) by specifying the sequence of decisions (N*365) over that horizon. This sequence constitutes a planning decision.

Is this the best solution?

To reply let’s consider the management only, i.e. let’s assume the reservoir has already been built.

Simplification: when the system is a periodic one, only 365 management decisions have to be defined.

taking decisions in full rationality

Cabora Bassa

irrigation

MOZAMBIQUE

Taking decisions in full rationality

at+1

st

ut

model

Decision:

volume of water to release every day from the dam in order to satisfy the farmers’ demand

the release plan
The release plan

It

?

catchment

at+1

st+1

ut

reservoir

+ users

m0 … m364

wt+1

slide7

s

*

t

The rule curve

It

catchment

a*t+1

ut

st+1

s*t+1

reservoir

+ users

m0 … m364

wt+1

slide8

s

s

*

s*

t

t

The rule curve

It

?

catchment

a*t+1

ut

s*t+1

reservoir

+ users

m0 … m364

wt+1

slide9

The rule curve

Rule curve for Cabora Bassa

Actual path

slide10

s

s

*

*

t

t

delay

The control policy

It

catchment

a*t+1

ut

mt(st)

s*t+1

reservoir

+ users

m0 … m364

wt+1

p= {mt(•) t = 0,1,…,h}

slide11

forecaster

â t+1

mt(st ,wt ,It ,at)

mt(st,wt)

delay

The control policy

It

catchment

delay

at+1

at+1

mt(st)

mt(st,wt,It,at)

mt(st ,wt ,ât+1)

st+1

ut

reservoir

+ users

m0 … m364

wt+1

delay

slide12

forecaster

â t+1

mt(st ,wt ,It ,at)

mt(st,wt)

delay

The control policy

Why a single decision ut? It’s more rational a whole set Mt !

It

catchment

delay

at+1

at+1

mt(st)

mt(st,wt,It,at)

mt(st ,wt ,ât+1)

st+1

ut

reservoir

+ users

m0 … m364

Mt

wt+1

delay

slide13

mt(st,wt)

delay

The control policy

It

catchment

delay

at+1

at+1

mt(st)

mt(st,wt,It,at)

mt(st ,wt ,at+1)

st+1

ut

reservoir

+ users

m0 … m364

wt+1

delay

simulation
Simulation

ANALYST

scenario choice

manag. policy

performance indexes

delay

delay

comparison & generation of policies

It

model of the manag. system

model of the physical system

catchment

at+1

delay

st+1

manag. policy

ut

reservoir + users

wt+1

set valued simulation
Set-valued simulation

ANALYST

scenario choice

performance indexes

DM

delay

delay

It

model of the manag. system

model of the physical system

catchment

at+1

delay

st+1

set valued manage policy

manag. policy

ut

Mt

reservoir + users

wt+1

in a deterministic world
In a deterministic world

Let’s introduce a simplification:

We are dealing with deterministic inflows

We know {a1,…,ah} for any time horizon {1,…,h}

single objective control problem

Defining criteria and

indicators

1. Reconnaissance

optimization

Identifying the model

Defining actions (measures)

*

*

history

B*mz. = utopia

3. Designing policy

p*mz.

MOZAMBIQUE

Single-Objective control problem

Design Procedure

Problem formulation

2. Conceptualisation

xt+1= ft(xt, ut, at+1)

at+1 ~ ft (•)

utUt (xt)

ut= mt(xt)

p = {mt(•) t = 0,1,…,h}

single objective control problem1

Defining criteria and

indicators

1. Reconnaissance

Identifying the model

3. Designing policy

Single-Objective control problem

Design Procedure

2. Conceptualisation

Defining actions (measures)

Integrated Modelling Framework

taking decisions in partial rationality

hydropower

Cabora Bassa

Cabora Bassa

Kafue

ZAMBIA

Kariba

irrigation

irrigation

ZIMBABWE

MOZAMBIQUE

MOZAMBIQUE

Taking decisions in partial rationality

Many interests

Many DMs

Full rationality

Partial rationality

xt+1= ft(xt, ut, at+1)

present situation
Present situation

BZim

BMozcon

BMoz

BZam

today

(BZamopt;BZimopt)

the optimal solution for mozambique
The optimal solution for Mozambique

BZim

E

D

F

BMozott

BMoz

BZam

DBMoz

utopia

today

(BZamott;BZimott)

BMozcon

the pareto frontier
The Pareto frontier

BZim

E

Pareto frontier

D

F

BMozott

BMoz

BZam

utopia

today

(BZamott;BZimott)

BMozcon

slide23

BZam

utopia

BZim

D BZam

alternative

today

BMoz

D BMoz

The Pareto frontier

multi objective control problem

zambia

at+1 ~ ft (•)

mozambique

utUt (xt)

ut= mt(xt)

zimbabwe

*

*

p = {mt(•) t = 0,1,…,h}

Multi-objective control problem

Formulation

Pareto frontier

xt+1= ft(xt, ut, at+1)

in an uncertain world
In an uncertain world

Considering the inflows as deterministic is an unrealitsic assumption.

However, we can not simply say that future inflows are unknow

Evaluation

Prediction

Rational decision

Predicting the future requires some past characteristic of the process to keep in the future:

modelling the inflow as a random process (stochastic).

THE STEADY STATE PARADIGM

decision making in uncertain condition example
Decision-making in uncertain condition - example

Knowing exactly what will happen, we would select alternative A2that returns 1500 €.

risk aversion
Risk aversion

Laplace criterion provide alternative A2 as the best choice.

And you, what would you select?

Maybe the worst case: min

partial rationality uncertain world the multi objective control problem

BZim

BMoz

BZam

Partial rationality + Uncertain worldThe Multi-Objective Control problem

Generating the whole Frontier is not always possible.

In some cases, interacting with the Stakeholders is more appropriate, thus generating the Front point by point. NEGOTIATIONS.

negotiations

BZim

afflussi

domanda irrigua

BMoz

BZam

Negotiations

40

35

30

25

20

15

10

5

0

BMoz

BZim

BZam

Just showing the value of the objectives could be not enough, in some cases showing the associated trajectories can be more useful ….

multi objective control problem1

Pareto frontier

Design Procedure

1. Reconnessaince

zambia

2. Conceptualisation

3. Policy design

4. Estimating the effects

mozambique

5. Evaluation

6. Comparison and negotiations

reasonable alternatives

zimbabwe

Mitigation

and compensation,

no

Agreement?

yes

Final decision

Multi-objective control problem

2. Conceptualisation

3. Designing policy

4. Estimating effects

5. Evaluation

6. Comparison and negotiations

Agreement?

slide31

Planning

TwoLe/P

MODSS

TwoLe

5. Evaluation

6. Comparison or negotiation

TwoLe/M

no

Mitigation and compensation

reasonable alternatives

Agreement?

yes

Management

Final (political) decision

Daily management

1. Reconnaissance

2. Conceptualisation

3. Designing alternatives

Stakeholders

4. Estimating effects

twole a 2 level modss
TwoLe: a 2 level MODSS

planning

Data

analyst

DM

stakeholders

TwoLe/P

models and policies

TwoLe/M

DM

users

management

release decision

operational control

ad