How does Artelys deliver optimization to your business?
It is a discipline involving a combination of skills in several areas: mathematics, decision-making, statistics, and computer science. It aims at determining an optimal solution, taking into account a set of constraints and variables inherent in the problem. In practice, there is no perfect mathematical solution to answer a problematic situation; however, Artelys ensures multiple ways to improve it.
What is optimization?
• Improvement
• Giving the best possible return; the scientific method to maximize or minimize an objective
• Determining the best value that can achieve a result, particularly when assessed according to several variables of interest and criteria of different nature (multicriteria optimization) that can be interconnected (cost and security, for ex.)
In the practices of the economic world, optimization is used often to increase profitability and / or reduce costs. Therefore, it is important to identify good solutions based on good models that are reliable and robust.
A concrete example to illustrate what can be achieved through optimization is: Each year the National Football League (NFL) establishes a 16 game schedule for the 32 teams that make up the league. Each team must play 8 “away” games in 17 weeks at various stadiums throughout the United States.
This schedule must take into account a large number of constraints and variables such as the availability of stadiums, the travel time between stadiums, the days of recovery between games, the broadcasting and many more...
These criteria are necessary to establish a balance between the profitability of the championship (mainly due to TV rights), the respect of fans, fair play competition, and balance between the different teams to avoid favoritism.
It is impossible to transcribe the problem in a perfect manner via the criteria and constraints. This is a discrepancy between the mathematical model* and the problem due to the difficulty of fitting a mathematical representation satisfactory to a particular problem. It makes the development of optimized solutions particularly challenging yet interesting. The NFLs’ head-honchos optimize the teams’ schedules to find a good and optimized solution that takes into account all the parameters of the problem yet there are other possibilities.
How to run a study in optimization?
The approach of a study can be structured in four stages: the diagram below is an example of the "bin packing".
In this example, we want to store seven boxes of different sizes into bins of equal size. The goal is to optimize the storage, that is to say, to store all the boxes in the minimum number of lockers possible.
A first step is to understand the problem and identify the objective and constraints*. This first phase requires strong technical expertise in optimization and in-depth knowledge of the problem to treat. In this context, optimization consultants can, in correspondence with their clients, make an analysis very useful, even essential to the success of the project. Those analyses will clearly identify the different actions to take and identify key constraints to be addressed.
The second step is purely technical and is to transcribe the problem in mathematical equations to obtain a mathematical model*. This work is done by optimization experts that partner-up with their clients to validate and test the model*.
The next step is to process the mathematical model* used by a computer program named Solver. Optimization problems are often tricky and deal with a large amount of data to process; they require powerful Solvers* and high performance hardware (computers, servers, etc.).
After validating the model* via empirical tests performed during its first utilization, the final step is to finalize the software design, to develop the necessary interfaces, and implement the solution within the existing system.
What is optimization software?
Optimization software is a tool used to solve a problem by giving it the ability to change parameters from one optimization possibility to another (constraints, systems to process, costs etc.)...
Artelys Energy Planner software facilitates the daily management of power plants.
Suppose that a production manager of a power plant seeks to satisfy his contractors at the least cost per energy demand daily. With this software you can quickly get an optimal production plan for the power plant and its contractors.
In case he wants to place orders on the spot market, he can measure the impact of these orders on the production plan and also develop a strategy for future orders.
Once he designates his orders on the spot market, he will be able to quickly update and adjust the production plan for a new optimal one. If an unexpected event occurs during the day for instance if one of its units fails, he may re-adjust and then re-optimize its production plan taking into account this new constraint*.
Finally if he wishes to conduct some strategic studies on the evolution of his power plant’s production, he can easily simulate different operational scenarios:
• If he had a new plant, to what extent would its daily production plan change?
• If one of his power plant needed maintenance, can he still meet the expected demand, including the demand for primary and secondary reserves?
Share this page
Linked in Viadeo Google Tweet this page FaceBook