Serdar i̇pli̇kçi̇ Optimization Applications in the Renewable Energy Systems

8/18/2019 Serdar ıplıkçı Optimization Applications in the Renewable Energy Systems

http://slidepdf.com/reader/full/serdar-iplikci-optimization-applications-in-the-renewable-energy-systems 1/33

The 4th Renewable Energy Systems Winter School │ Prof. Dr. Serdar İPLİKÇİ

OPTIMIZATION in RENEWABLE

ENERGY SYSTEMS

OPTIMIZATION in RENEWABLE ENERGY SYSTEMS

1/3316/01/2015

Prof. Dr. Serdar İPLİK Çİ

Pamukkale University, Dept. of Electrical and Electronics Eng., Kınıklı Campus, 20040, Denizli

e-mail: [email protected]

web: www.pau.edu.tr/iplikci

phone: +90 (258) 2963197





2/3316/01/2015

CONTENTS1. Introduction

2. Optimization Techniques1. Introduction

2. Heuristic Optimization

3. Gradient-Based Optimization

3. Optimization Applications in the Renewable and

Sustainable Energy Systems1. Wind Power

2. Solar Energy

3. Geothermal Energy

4. Bioenergy

5. Hybrid Systems

4. Modeling and Prediction

5. Conclusions





3/3316/01/2015

1) Introduction1.1) Why Optimization?

• Due to the increasing demand and limited sources worldwide, sustainability is of great

importance

• It is necessary to develope efficient methods that allow us to produce and consume

the energy very effectively.

• The extension of energy sources and the information structure allow a fine screening

of energy resources, but also require the development of tools for the analysis and

understanding of huge datasets about the energy grid.• Key technologies in future ecological, economical and reliable energy systems are

• Energy prediction of renewable resources

• Prediction and monitoring of energy consumption

• Efficient planning and control strategies for network stability

• To enable ecologically and financially feasible projects, optimization methods have

taken over a key role for planning, optimizing and forecasting sustainable systems.





4/3316/01/2015

1) Introduction1.2) Optimization in the Literature

Source: Web of Science





5/3316/01/2015

2) Optimization Techniques2.1) Introduction

2.1.1) General Format and Short Notation

Objective Function Design Variables

min ()s . t . : () 0 ∈ ℰ

() ≥ 0 ∈ ℐ

min,,…,(, , … , )

s. t: (, , … , ) 0 ∈ ℰ(, , … , ) ≥ 0 ∈ ℐ

: ℝ ⟼ ℝ ∈ ℝ 1,2, … ,

⋮

()

Local minimum

Global minimum





6/3316/01/2015

2) Optimization Techniques2.2) Heuristic Optimization

2.2.1) Introduction

•

Heuristic methods are the methods that produce sufficient (even if not optimum)solutions to the large scale problems very rapidly.

• Meta-heuristics are generalizations of heuristics in the sense that they can be applied

to a wide set of problems.

• Heuristic methods can be categorized as follows:• Trajectory vs population

• memory-based vs memoryless

• nature-inspired vs non-nature-inspired


http://slidepdf.com/reader/full/serdar-iplikci-optimization-applications-in-the-renewable-energy-systems 7/33The 4th Renewable Energy Systems Winter School │ Prof. Dr. Serdar İPLİKÇİ


7/3316/01/2015


2.2.2) Trajectory Methods

Trajectory methods start the search process with a single solution. During the search

process, the solution is updated iteratively and the thus the outcome is also a single

optimized solution. Most of them are extensions of simple iterative improvement

procedures that incorporate techniques that enable the algorithm to escape from local

optima. Some of the trajectory methods are:

•

Hill Climbling (HC)• Simulated Annealing (SA)

• Tabu Search (TS)

• Greedy Randomized Adaptive Search Procedures (GRASP)

• Variable Neighborhood Search (VNS)

Some of the meta-heuristics trajectory methods are :

• Iterated Local Search (ILS)

• Pareto Archived Evolution Strategy (PAES)

• Multi-Objective Simulated Annealing (MOSA)




8/3316/01/2015


2.2.3) Population-based Meta-Heuristics Methods

Population-based heuristics use a population of solutions which evolve during a given number of

iterations, also returning a population of solutions when the stop condition is fulfilled. Themain population-based heuristics include:

• Genetic Algorithms (GA)

• Evolutionary Algorithms (EA)

• Scatter Search (SS)

• Path Relinking (PR)

• Memetic Algorithms (MA)

• Ant Colony Optimization (ACO)

• Particle Swarm Optimization (PSO)

• Estimation of Distribution Algorithm (EDA)

• Differential Evolution (DE)

• Artificial Bee Colony Optimization (ABCO)

Population-based meta-heuristics are:• The Multiobjective Tabu Search (MOTS)

• Non-dominated Sorting Genetic Algorithm (NSGA/NSGA-II)

• Pareto Simulated Annealing (PSA)

• Single Front Genetic Algorithm (SFGA)

• Strength Pareto Evolutionary Algorithm (SPEA/SPEA-II)

• Pareto Envelope-based Selection Algorithm (PESA/PESA-II).




9/3316/01/2015


2.2.4) Genetic Algorithm Cycle

Table of fitness values

N F F F ,,,

21 calculate

Sort from smallest to highest

Population

selection

crossover

roulette wheel

1 0 0 1 1 1 0 1 0 0 1 0 1 1 0 1

1 1 0 0 1 0 1 1 0 1 1 1 0 1 0 0

Parent 1

Parent 2

Offspr ing1

Offspr ing2

1 0 0 1 1 1 0 1 0 1 1 1 0 1 0 0

0 0 1 0 1 1 0 11 1 0 0 1 0 1 1

2X

N X

3X

1X

mutation

1 0 1 1 1 0 1 0 0 1 0 1 1 0 1or iginal

mutated 1 0 0 0 1 1 0 0 0 1 1 1 1 1 0 0

GENETIC ALGORITHM CYCLE

fitness function

number of population

fitness of the i th solution

i th individual (solution)

binary counterpart of the

i th solution

0

(, , … , )

N

(, , … , )

decimal2binary

⋮

OPTIMIZATION i RENEWABLE ENERGY SYSTEMS




10/3316/01/2015

2) Optimization Techniques2.3) Gradient-Based Optimization

2.3.1) Mathematical Basics

, , … ,

, , … ,

, , … ,

⋮ , , … ,

Gradient Vector

, , … ,

, , … ,

, , … ,

⋯ , , … ,

, , … ,

, , … ,

⋯ , , … ,

⋮ ⋮ ⋱ ⋮

, , … ,

, , … ,

⋯ , , … ,

Hessian Matrix

, , … ,

, , … ,

, , … ,

⋯ , , … ,

, , … ,

, , … ,

⋯ , , … ,

⋮ ⋮ ⋱ ⋮ , , … ,

, , … ,

⋯ , , … ,

Jacobian Matrix





11/3316/01/2015


2.3.2) Taylor Expansion, Descent Direction and Optimality Conditions

Taylor Expansion

∆ ∆ 12 ∆ ∆ ℎ. .

Descent Direction

First-order Optimality Conditions: ∗

Optimality Conditions

Update Rule

If one can find a search direction that satisfies <

+ ← kuralı ile güncelleme yapılır.

Second-order Optimality Conditions : ∗ positive definite

≅

< →

< 0





12/3316/01/2015


2.3.3) First- and Second-order Methods

Steepest Descent:

Conjugate Gradient: −

Newton: −

Modified Newton: −

First-order Methods

Second-order Methods





13/3316/01/2015


2.3.4) Quasi-Newton Methods and Second-order Approximate Methods

Davidon-Fletcher-Powell (DFP) :

, + ∆ ∆ ∆

, +

Gauss-Newton (GN):

−

Levenberg-Marquardt (LM): − .

Quasi-Newton Methods

Second-order Approximate Methods

Broydon-Fletcher-Goldfarb-Shanno (BFGS) :

− , + ∆ , +





14/3316/01/2015

3) Optimization App. in the Renewable and Sustainable Energy Systems

PROBLEM ÇÖZÜM

Community-scale renewable energy systems planning is an

important problem consisting of justifying the allocation

patterns of energy resources and services, formulation of localpolicies regarding energy consumption, economic

development and energy structure, and analysis of

interactions among economic cost, system reliability and

energy-supply security.

• Interval Linear Programming (ILP)

• Chance-Constrained Programming

• Mixed Integer-Linear Programming (MILP)

a long-term dynamic multi-objective planning model for

distribution network expansion along with distributed energy

options

Immune Genetic Algorithm (I-GA)

minimum cost expansion of power transmission networks

under carbon emission trading programs

• Mixed-Integer Programming (MIP)

• Genetic Algorithms (GA)

• Simulated Annealing (SA )

• Tabu Search (TS)

annual peak load forecasting in an electrical power system

with the aim of minimizing the error associated with the

estimated model parameters

Particle Swarm Optimization (PSO)

new renewable energy sources penetration

and congestion management so that electricity supply and

demand are always evenly balanced

• Nelder–Mead Simplex (NMS) and PSO

• Honey Bee Mating Optimization (HBMO)

• Ant Colony Optimization (ACO), ANN, GA

Energy demand prediction • Yapay Sinir Ağları (ANN)

• Destek Vektör Makineleri (SVM)





15/3316/01/2015

3) Optimization App. in the Renewable and Sustainable Energy Systems3.1) Wind Power

3.3.1) Introduction

•

Wind is a periodical phenomenon for large geographical areas like Mexico.• the increasing sizes of turbines and the lower prices per installed production

capacity of electricity.

• Wind energy systems may not be technically viable in all locations because of

low wind speeds and the fact that it is more unpredictable than solar energy

• Areas where winds are stronger and more constant, such as offshore and high

altitude sites, are preferred locations for wind farms.

•

An accurate estimation of wind speed distribution

the site selection of windfarms

• Bayesian modelmodeling long-term wind speed distributions





16/3316/01/2015


3.3.1) Optimization in Wind Turbine Design

• In particular, two important problems are often considered: wind turbine and wind farm layout

The power output of a turbine is a function of the:• density of the air

• area swept out by the turbine blades

• cube of the wind speed

• Numerous metrics are used to measure the power quality of a wind turbine, such as

• the power factor, reactive power,

• Harmonic distortion

the optimization of the geometrical parameters of the rotor

configuration of stall-regulated horizontal-axis wind turbines with

the aim of achieving the best trade-off performance between the

total energy production per square meter of wind park and cost

Multi-Objective Evolutionary Algorithm

(MOEA)

the optimization of the ranges of gearbox ratios and power ratings

of multihybrid permanent-magnet wind generator systems

Genetic Algorithms (GA)

determining the optimum capacity taking into account uncertainties

arising from wind speed distribution and power–speed

characteristics

Mixed-Integer Nonlinear Programming

(MINLP)

prediction of wind speed at a selected location based on the data

collected at the neighbouring locations

Fuzzy Logic Modelling





17/3316/01/2015


3.3.2) Optimization in Wind Farm Layout

• Wind farm layout consists of determining the optimum positions of wind turbines

within the farm in order to maximize energy production.

optimal placement of wind turbines for maximum production capacity while

limiting the number of turbines installed and the acreage of land occupied by

each wind farm


optimum wind farm configuration problem which is driven by an integralwind farm cost model based on the cumulative net cash flow value

throughout the wind farm’s lifespan

Evolutionary Algorithms (EA)

wind turbine placement based on wind distribution with the aim of both

maximizing the wind energy capture and minimizing an index that

determines constraint violations

Multi-Objective Evolutionary

Algorithm (MOEA)

determining the optimal type, number and placement of wind turbines

considering the given wind conditions and wind park area

Mixed-Integer Nonlinear

Programming (MINLP)






18/3316/01/2015

3) Optimization App. in the Renewable and Sustainable Energy Systems3.2) Solar Energy

3.2.1) Introduction

• Solar energy is radiant energy that is produced by the sun. In many parts of the world,

direct solar radiation is considered to be one of the best prospective sources of energy.

• The main ways to convert solar radiation into energy are active and passive solar

design.

• Passive solar design is often based on the optimal design of buildings that capture the

sun’s energy in order to reduce the need for artificial light and heating. Regarding

passive solar systems, a primary interest for researchers in solar energy is related to

the design and optimization of solar energy homes.

• Active solar design is based on water heating converting solar radiation into heat using

photovoltaic panels and solar cells to convert the solar radiation into energy.

•

In order to design both active and passive solar energy systems, radiation data areneeded for the studied location.

calculating solar radiation levels over complex mountain terrains using data from only

one radiometric station

ANN, Neuro-Fuzzy






19/3316/01/2015

3) Optimization App. in the Renewable and Sustainable Energy Systems3.2) Solar Energy

3.2.2) Active Solar Design

• An interesting problem related to photovoltaic systems is the optimal determination of

their size. The sizing optimization of a stand-alone photovoltaic system is a complex

optimization problem which aims to obtain acceptable energy and economic cost for

the consumer, and a relatively correct energy supply quality.

identifying the electrical parameters of photovoltaic solar cells and modules to

determine the corresponding maximum power point from the illuminated current –

voltage characteristic

Genetic Algorithms

(GA)

maximizing the thermal performance of flat plate solar air heaters by considering

the different system and operating parameters

Genetic Algorithms

(GA)

determining the tilt angle of photovoltaic modules with the aim of maximizing the

electrical energy output of the modules

Particle Swarm

Optimization (PSO)

Optimal sizing of photovoltaic systems ANN ve GA






20/3316/01/2015

3) Optimization App. in the Renewable and Sustainable Energy Systems3.3) Bioenergy

3.3.1) Introduction

• Bioenergy is renewable energy made available from materials derived from biological

sources. Biomass, a renewable energy source, is biological material from living, orrecently living organisms, including plants and animals.

• Biomass is one of the most promising renewable energy sources, but more research is

required to prove that power generation from biomass is both technically and

economically viable

• Biomass can be burned to produce steam for making electricity, or to provide heat to

industries and homes.• In addition biomass can be converted to other usable forms like methane gas, ethanol

fuel and biodiesel fuel.

• Biomass power plants exist in over 50 countries around the world and supply a

growing share of electricity.

• The sustainability of electricity generation from biomass must be assessed according

to the key indicators of price, efficiency, greenhouse gas emissions, availability,

limitations, land use, water use and social impacts.

Finding optimal location of biomass-fuelled systems for distributed power generation

with forest residues as biomass source

binary PSO-based

method






21/3316/01/2015

3) Optimization App. in the Renewable and Sustainable Energy Systems3.4) Geothermal Energy

3.4.1) Introduction

• Geothermal energy is the energy contained as heat inside the Earth. Geothermal heat

pumps are a highly efficient, renewable energy technology for heating and cooling.

• This technology relies on the fact that, at depth, the Earth has a relatively constant

temperature, warmer than the air in winter and cooler than the air in summer.

• The main advantage of using geothermal energy is that this renewable energy source

can provide power 24 h a day due to it is constant, without intermittence problems

compared to other renewable resources such as wind or solar energy.

• It is expensive to build a power station but operating costs are low, resulting in low

energy costs for suitable sites.

• Geothermal power plants now exist in 19 countries, and new plants are commissioned

annually. However, only a small fraction of the geothermal potential has beendeveloped so far, and there is ample space for an accelerated use of geothermal energy

both for electricity generation and direct applications.

optimization of the exploitation system of a low enthalpy geothermal aquifer, with

the aim of determining the annual pumping cost of the required flow and the

amortization cost of the pipe network, which carries the hot water from the wells

to a central water tank, situated on the border of the geothermal field







22/3316/01/2015

3) Optimization App. in the Renewable and Sustainable Energy Systems3.5) Hybrid Systems

3.5.1) Introduction

• In the last decade, there has been a spectacular increase the interest in optimizing the design and

control of stand-alone hybrid power generation systems in order to manage energy between themaximum energy captured and consumed energy.

• The aim of optimizing the mix of the renewable system is maximizing its contribution to the peak

load, while minimizing the combined intermittence at a minimum cost.

the economic environmental dispatching of a hybrid power system

including wind and solar thermal energies.


Algorithm (MOEA) and GA

the multi-objective design of isolated hybrid systems where the

objectives to minimize are the total cost throughout the useful life of

the installation and the pollutant emissions


Algorithm (MOEA) and Strength Pareto

Evolutionary Algorithm (SPEA)

Optimal sizing a hybrid solar–wind-battery system with the aim of

minimizing the annualized cost system and the loss of power supply

probability


Solving the wind-photovoltaic capacity coordination for a time-of-use

rate industrial user with the aim of maximizing the economic benefitsof investing in a wind generation system and a photovoltaic

generation system

Particle Swarm Optimization (PSO)

Optimal sizing of stand-alone photovoltaic-wind generator systems,

which selects the optimal number and type of units to minimize the

cost subject to the constraint that the load energy requirements are

completely covered





The 4th Renewable Energy Systems Winter School │ Prof. Dr. Serdar İPLİKÇİ 23/3316/01/2015

4) Modeling and Prediction4.1) Introduction

4.1.1) The Concept of Modeling

• Modeling is a method that is used when a real-world «process» cannot be described

analytically or mathematically. Here, the «process» can be all sorts of biological,physical, chemical, electrical, mechanical, meteorological, social and financial

dynamical systems.

• There are many such systems in the real world. For instance, dynamics of weather

conditions is so complex that it cannot be described mathematically, and thus it should

be modeled.similarly, stock market is a very complex system that incorporates manyvariables and parameters. There some problems in the energy systems that need

modeling:

• Electrical load prediction

• Energy demand prediction

• Therefore, when needed, we have to collect sufficient data from such energy systems

and then try to obtain a reliable model.






4.1.2) Data Types

Input-Output Data

⋮

⋮

MM Input Data Output Data

... ...

1 ... ...

2

...

...

⋮ ⋮ ⋮

N ... ...

∈ ℝ ∈ ℝ

unknown system






4.1.2) Data Types Time Series

1 2 3 4 5 6

… 1

?

Input Data Output Data

...

1

1

2 ...

1

2 2 3 ... 1 2

⋮ ⋮ ⋮ 1 ... 1





4) Modeling and Prediction4.2) Modeling and Model Selection

4.2.1) Generalization, Approximation and Prediction Errors

MM Input Data Output Data Model Output

... ... ...

1 ... ... ...

2 ... ... ...

⋮ ⋮ ⋮ ⋮ N ... ... ...

∈ ℝ ∈ ℝ

MM , =

⋮

⋮

real model

optimum model

HYPOTHESIS SPACE

predicted model

TARGET SPACE






4.2.1) Emprical Error, Overfitting and Splitting Data

ℎ

Over-learning Under-learning

model complexityEmprical error

training

validation

test






4.2.2) Objective Function

min , ,

∈

=

∈

=

1

, ,

∈VL

=

Complexity

best model





4) Modeling and Prediction4.3) Example: Chaotic Time Series Predictyion by Artificial Neural Nets

4.3.1) Architecture of an Artificial Neural Network

Σ

1

ℎ

Σ

1

ℎ

Σ

−

1

ℎ

Σ

1

ℎ

⋮ Σ

ç

1

,ç

,ç

,−ç

,ç

⋮ ⋮

−

,

,

ℎ −

−

,

⋮,

,

⋮,

⋮

,

⋮,

⋮

,ç

⋮,ç

ç

+ ←

1

ℎ

ç

Σ

i n p

u t s i g n a l s

output signal

⋮

input weight

summing block

bias weight

activation

function

output weight

input weight

Single neuron

Neural network






4.3.2) Data

0 20 40 60 80 100 120 140 160 180 2000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

k

x k

Chaotic Logistic Map + 3.9 1 , 0.2






4.3.3) Model Output

0 50 100 150 200 2500

0.2

0.4

0.6

0.8

1LM VERI:logistic EGITIM VERISI SAYISI=119 TEST VERISI SAYISI=80 NORON SAYISI=10 GURULTU GENLIGI=0.01

gercek yorunge

en iyi model cikisi

tahmin cikisi

50 100 150 200 250

0

0.05

0.1

0.15

0.2

iterasyon

h a t a

egitim hatasi

training error

test error

190 192 194 196 198 200 202 204 2060

0.2

0.4

0.6

0.8





5) Conclusions

• The optimization methods that have beenn used for solving optimization problems inrenewable energy systems have been developed day-by-day. They are especially used

in hybrid systems.

• Some of these methods are based on the traditional methods like mixed-integer,interval linear-programming, Lagrangian relaxation, quadratic programming, andNelder–Mead Simplex, while others are based on some heuristics methods such as GAand PSO.

•

On the other hand, the multi-objective function problems in the energy systems havebeen solved by Pareto-optimization techniques.

• The problems in the renewable energy systems that can be solved by optimizationtechniques are• Planning• Management of supply-demand balance• Optimization of design parameters

• Prediction of power curve• Configuration• Optimization of economical load distribution• Wind-photovoltaic capacity coordination• Modeling and prediction




THANKS...

Date post:	07-Jul-2018
Category:	Documents
Upload:	angarali26
View:	215 times
Download:	0 times

Serdar i̇pli̇kçi̇ Optimization Applications in the Renewable Energy Systems

Documents