My name is Batur Usta and I currently work at Tupras as a Modeling and Optimization Engineer. My work consists of production optimization using linear programming where we attempt to come up with the most profitable production schema for the month. Since refining is a continious process and the product quality is highly dependent on feed quality which unfortunately is ever changing; achieving the most feasible solution is a difficult task. In this regard, we use a set of package programs specialized for refinery optimization. The whole process begins with analyzing field data to update our models, which is currently being done manually using spreadsheet programs. My aim is to do it more time efficently and accurately with the help of programming languages such as R and Python. By using linear programming algorithms and possibly machine learning, I believe I will be able to reduce my workload and increase my efficency.
Constrained optimization is where the system has various limitations, or constraints which are required to be met when achieving optimum results. In linear programming, constraints are written in linear form only, for example, x+2y > 4 is a linear constraint where the variables x and y are limited. A typical linear programming problem has multiple constraints. In R, constrOptim program can be used to solve such systems. The dimensions and constraints are presented in matrix form and the program is fed an initial point which must be in the solution zone. The function normally finds the minimum solution, but with an additional command it can find the maximum solution too.
Oil market is a volatile, sensitive and unpredictable market. It is maybe the most important commodity of our world and oil prices and the potential drastic changes in oil price can affect the economy of most countries. This work attempted to achieve a reliable forecasting method for Brent oil prices using R programming using various methods and comparing the results. Brent oil is one of the index oils where the prices of all oils are derived from. The results concluded that the oil market is indeed unpredictable and coming up with a solid forecasting scheme is challenging.
One of the strongest aspects of R is its data visualization capabilities. Thanks to libraries such as ggplot2, Rcolorbrewer, plotly etc. data visualization is easily and effectively done in R. It is also possible for R to create interactive charts where you can see details of the data on the graph. Interactive graphs and dynamic graphs strengthens our plots, efficent use of such tools will make our projects stand out.