ENERGY MODELLING USING AIMMS

Prof Nnamdi Nwulu (PhD, PrEng, SMIEEE, SMSAIEE)

 CONTENTS
• What is a mathematical model?
• What makes a good mathematical model?
• What are the modelling pitfalls to avoid?
• What model should I use?
• Steps in building a mathematical model?
• Types of mathematical models?
• What are probabilistic models?
• What is mathematical optimization?
• A practical example of mathematical optimization
• What software’s are useful?
• Conclusions
 What is a mathematical
• Amodel?
mathematical model is a description of a system using mathematical
concepts and language.
• Anything that can be expressed in terms of mathematical formulae
and that is amenable to mathematical analysis
• The process of developing a mathematical model is termed
mathematical modelling.
• A model may help to explain a system and to study the effects of
different components, and to make predictions about behaviour.
• Mathematical models are used in
* the natural sciences (such as physics, biology, earth science, meteorology)
* engineering disciplines (e.g. computer science, artificial intelligence)
* the social sciences (such as economics, psychology, sociology and political
science)
 What is a mathematical model
cont’d?
To summarize:
• Mapping real world entities/mechanisms/phenomenon to
mathematical abstractions.
• Constructing and developing the model with mathematical theories.
• Reasoning in pure math: Definitions ! Theorems ! More
theorems...
• Discovery by other means, especially simulation: computer simulation
of math models
• Solving real-world problems within the theoretical framework that
you developed (proposing a solution that proves to work based on
your theory)
 What makes a good
mathematical
• Reasonable: model?
reasonable mapping between the real and the math
• Formal: formal way of developing theories:
* Mathematically sound/rigorous.
* New insight (Don’t just restate the question in math without any new insight)
• Generalizable: working on a general framework rather than too
specific
* Can I solve the problem if the occasion changes?
* What else can I do/learn with this model?
• Simplicity: choose the simplest one that can solve your problem!
* Do not be obsessed with fancy stuff.
• Verifiability: how do you know your model is not wrong?
* Does your solution solve your problem and is it compliant with data?
* Other supportive theories/conclusions.
 What are the modelling pitfalls
to model
• Your avoid?
is not a “math model” at all!
— You did not find the right math tool to address the problem.
— You did not well define the stuff mathematically (dirty mapping).
• Your model and solution are unrelated!
— Your math model must explain why you propose the solution and why your solution works.
• Your model still cannot solve the problem!
— Model too complex or you do not know how to solve.
• You have one model for Scenario 1 and another for Scenario 2!
— Model too specific, not generalizable.
• You have a perfect model that can explain everything!,
— Your model must be VERIFIABLE!!
 What model should I use?
• What real-world problem do you have?
• What do you have in your theoretical arsenal?
• Does the model capture the most important mechanism of the
problem? (white/black boxes)
• Can you use the model to solve the problem?
 Steps in building a
The mathematical model?
standard way to build a mathematical model then involves the following
four steps:
1. Formulate the modelling problem: What is the question to be answered?
What type of model is appropriate to answer the question?
2. Outline the model: Which effects should be included in the model, which
are negligible? Write down relations between the states.
3. Practicability check: Is the model “solvable”, either by analytical methods
or numerical simulation? Do I have access to all the parameters in the
model? Can the model be used to make predictions?
4. Reality check: Make predictions of known phenomena and compare with
available data (qualitatively or quantitatively). Note that there is a trade-off
between simplicity of a model (easier to analyze and interpret) and
“realism” or accuracy of the model (potentially more complicated, analysis
may be through computer simulations only).
 Types of mathematical
models?
• deterministic ODE models (e.g. population growth, chemical reactions
mechanical systems and analogues, climate, . . . )
• stochastic models (e.g. growth of small populations, chemical
reactions in cells, asset prices, weather prediction, . . . )
• optimization models (e.g. minimization of energy consumption, etc, . .
.)
• discrete or continuous flow models (e.g. queuing problems, traffic,
logistics, load balancing in parallel computers, . . . )
• statistical models (e.g. distribution of votes, change in the
precipitation rate, wage justice, criminal statistics, Google
PageRank, . . . )
• Combination of the aforementioned modelling approaches yields a
hybrid model
 What are probabilistic
models?
• Stochastic processes :
• Markov Chains, Hidden Markov Model (HMM) e.g. application in disease/information
spreading
• Poisson process, branching process e.g. queueing, population growth
• Machine/Statistical learning & Pattern recognition
• Supervised learning (logistic regression, neural networks, etc.) e.g. hand-writing
recognition and other common classification tasks
• Unsupervised learning e.g. clustering, dimensionality reduction
• Concerns with probabilistic models:
• Parameter estimation from data (training/learning)
 What is mathematical
optimization?
• This is also called programming.
• Linear programming, integer programming and their solvers
• Basic non-linear programming
• Optimization methods
* With gradient and even Hessian
* General heuristic searching algorithms (if too hard to solve) e.g. simulated
annealing, genetic algorithms
• Concerns with Programming/Planning:
• How to formulate your decision variables?
• How to define the target/reward function?
• Solver available or we have to do with a heuristic search?
• Then how large is your search space?
Optimization Definition
• Optimization is used to find the “best” value
• “Best” defined by us, the analysts and designers
• Constrained opt  Linear programming
• Linear constraints
• Complicates the problem
• Some binding, some non-binding
• Visualize via a ‘feasible region’

12
Formulating the Problem
• Objective function
• Constraints
• Decision variables
• Variable bounds
• Standard form
• min c (x)
• s.t. Ax = b
xmin <= x <= xmax

13
 What software’s are useful?
• MATLAB — large number of discipline specific toolboxes
• MATHEMATICA — useful for analytical solutions
• WEKA — free data mining software
• AIMMS — free for academic use.
• AMPL — free student license.
• APMonitor — free for academic or commercial use with interfaces to
Python and MATLAB.
• ASTOS — free for academic use.
• CPLEX — free for academic use.
• Galahad library — free for academic use.
• GUROBI — free for academic use.
• LIONsolver — free license for academic use.
 What software’s are useful
cont’d?
• MIDACO Global optimization software based on evolutionary
computing, MINLP, Parallelization (Excel, Java, Matlab, Python, C/C++,
R and Fortran)
• MINTO — integer programming solver using branch and bound
algorithm. Freeware for personal use.
• MOSEK — free for academic use.
• OptimJ — Java based modeling language. Free Edition includes
support for lp_solve, GLPK and LP or MPS file formats
• PottersWheel — Parameter estimation in ordinary differential
equations (Free Matlab toolbox for academic use).
• SCIP — Free only for research purpose as a member of a non-
commercial and academic institution.
• WORHP — free of charge for academic use.
• XPRESS — free for academic use.
 Some Tools
General Mathematical Programming Software:
LP_Solve [http://lpsolve.sourceforge.net/5.5/IDE.htm]
Lingo [http://www.lindo.com/]
GAMS[http://www.gams.com/]
AIMMS [http://www.aimms.com/]
THANK
YOU!!