Araştırma Alanları


research_L1001898_500px

ŞEHİR, Mühendislik ve Doğa Bilimleri alanlarında güçlü bir araştırma potansiyeline sahiptir. Endüstri Mühendisliği Programı kapsamındaki araştırmalar, bu amaçla endüstrinin farklı alanlarına yönelmiştir. Öğretim üyelerimizin araştırma gruplarına ve araştırma alanlarına göz atabilirsiniz.

Modeling and Optimization of Electricity Generation/Transmission/Distribution Systems

The primary purpose of electricity power system is to provide a reliable and economic supply of electricity.   Since the electricity demand is increasing and the components in the existing system are aging, the electric power system must be expanded and modernized.  Generation expansion planning problems are solved to determine what, when and where to built the new technologies to meet the future demand over long term planning horizon.  Due to the increasing awareness for clean air and global warming, the power grid should also be designed to be environmental friendly. 

An electric power system consists of mainly three major components which are (i) the electricity generation system, (ii) transmission lines which consists of high voltage lines and connect the generation unit to substations, and (iii) distribution lines which consist of lower voltage lines and connect the substations to the final customers.  A simple representation of a power system can be given as below.

The power system traditionally uses large-scale centralized generation units relying on technologies such as nuclear or coal burning.  However, the power system can also include distributed generation units which are small scale generation units located close to the final consumers.  Some of the technologies available for the distributed generation are fuel cells, micro-turbines, photovoltaic systems, small wind turbines and so on.  Some of the distributed generation units have the co-generation capability. Co-generation means that they can use the generated heat together with the electricity.

There are also technological developments which may be used to construct a more reliable, effective and energy efficient power grid.  These developments are called as Smart Grid Technologies. There are mainly five key Smart Grid Technologies.

One of the main objectives of Smarter Grid is provide real time information to all the utilities in the power system.  In order to provide this, it is required to have technologies providing two-way communication.  These technologies are called as “integrated communication technologies”, and they will support to integrate smart sensors, control devices and other intelligent technologies into grid.  In order to increase the reliability and efficiency of the grid, it is required to know the state of the system.  Therefore, new digital technologies are developed; these are called “sensing and measurement” technologies.  These technologies collect the data and transfer the data to be analyzed by using two-way communication system.  By means of such technologies, it is possible to get the information about the state of the grid component and overall system, provide outage detection and response; enable utilizing demand response programs and so on.   There are many developments in new materials technologies, nanotechnologies, advanced digital designs and so on.  In order to obtain modern grid, it is necessary to consider these technologies as an available technologies for expansion.  The general name for this group of technologies is “advanced grid components.”  Some of the advanced technologies are superconducting transmission cable, fault current limiters, advanced energy storage, distributed generation, advanced transformers and so on.  Taking good decisions in a short time is very important for operating the power grid. Therefore, there are some new technologies, called “Decision Support and Human Interfaces” technologies, to convert the complex power system data into information which can be understood by the operator easily.  Some of them are visualization tools and systems, operator decision support systems (what-if tolls, alerting tools, etc.), semi-autonomous agent software, real-time dynamic simulator training and so on.   Another group of technologies required to construct a Smart Grid are called “advanced control methods (ACM).”  ACM technologies are the devices and algorithms that will monitor the essential grid components.  They collect data (Sensing and Measurement), analyze the data and provide rapid diagnosis (Improved Interfaces and Decision Support), determine and take automated or provide appropriate response (Integrated communications, Advanced Components).  Therefore, ACM technologies rely on and contribute to each of the four key technology areas.

 

Table below lists possible objectives considered in generation expansion planning problems; decision variables used and their definitions; some constraints commonly used and stochastic parameters which can be part of the problems.

Ekran Resmi 2015-06-26 16.03.10

 

I am interested in developing mathematical models to solve electricity generation expansion planning problems where important problem objectives, such as cost, greenhouse gas and pollutant emissions, reliability and risk are explicitly considered under an uncertain environment.

I am also interested in developing methods to incorporate the effects of the Smart Grid technologies described above in to the mathematical model.  Since energy produced by renewable energy sources is depends on the uncontrollable source such as wind and sun lights, they are called intermittent or non-dispatchable technologies.  Therefore, there are uncertainties associated with renewable energy sources and I am also interested in developing methods to introduce these uncertainties into the generation expansion planning problem.

 

System Reliability Optimization

Reliability is the probability that an item (or a system) will perform a required function under stated conditions for a stated period of time.   Redundancy allocation problems are solved to determine the configuration of the system and types of the components to be used in the system so that some objectives are optimized under some system related constraints.

I am particularly interested in redundancy allocation problems for a series-parallel system.  In as series-parallel system, there are m subsystems which are each configured as a parallel structure and those subsystems are connected to each other in series. Figure below presents a typical series-parallel system. For each subsystem, there are multiple component choices available and the problem is the determination of the component choice and redundancy level for each subsystem.

Most system reliability optimization studies assume that the reliability values of the components are deterministic, i.e., known with certainty. However, in practice the reliability of a component is generally estimated from field or test data, and therefore, there is some uncertainty associated with the estimates. Both reliability estimates and time-to-failure can also be considered as random variables. Decision-makers clearly prefer solutions with lower estimation uncertainty.  Therefore, when it is ignored, undesirable solutions can be obtained which have unacceptable level of uncertainty and risk.   Therefore, I am interested in developing mathematical models and algorithms to maximize the reliability of the system as well as to minimize the variance.

 

Airline Planning

The airline industry encounters many optimization problems.   Firstly, an airline company determines the flights to be flown in a given time period. This problem is called flight scheduling. Then, fleet assignment problem is solved to allocate available aircrafts to flight legs according to the estimated demand for each flight leg. Third step is aircraft routing problem, which guarantees that each aircraft receives adequate maintenance checks. Finally, the crew scheduling problem is solved.  The crew scheduling problem is usually divided into two separate problems, namely the crew pairing and the crew assignment problems.   The crew pairing problem aims at finding the least costly subset of pairings, which cover the scheduled flights.  A crew pairing is defined as the sequence of flight legs which starts and ends at the same crew base.  It is possible to have overnight rests between the flights. Then, the crew assignment problem is solved to allocate the pairings to the individual crew members.

Another important issue is that the airline companies face with disruptions during the operations.  These disruptions due to weather conditions, maintenance problems, adding an extra flight to the regular flight schedule and so on are leading to higher operational crew cost in practice. These kinds of disruptions might result in delaying or canceling some scheduled flights.  These disruptions in the crew pairing schedule might be considered at the operational level.  Then, a crew recovery problem is solved to restore the disrupted schedule. The objective of this problem is to modify the disrupted schedule as quickly as possible with minimum costly reassignment.   However, it is also possible to build a robust schedule against those disruptions by considering possible disruptions at the planning level.  This type of problems is called robust crew pairing.

Although I only worked on robust crew pairing problem when the disruption is to add an extra flight to predetermined (regular) flight schedule, I am interested in all the optimization problems faced in the airline planning described above.

MDP and POMDP-Based Approaches to Stochastic Shortest Paths with Dynamic Learning

We consider a probabilistic path planning problem where an agent needs to quickly navigate from one given location to another through an arrangement of arbitrarily-shaped regions which are possible obstacles. At the outset, the agent is given the respective probabilities that the regions are truly obstacles and, when situated on a region’s boundary, the agent has the option to disambiguate it, i.e., learn at a cost if it is truly an obstacle. It is assumed that there is a pre-specified limit on the number of available disambiguations. The central question is to find an algorithm that decides what and where to disambiguate en route so as to minimize the expected length of the traversal from start to destination. This problem is called the Stochastic Shortest Path with Dynamic Learning Problem (SDL). The SDL problem and many of its variants have been shown to be NP-hard.

The objectives of this research are as follows: (1) investigate approximation schemes for stochastic dynamic programming and Markov decision process (MDP) frameworks for the SDL problem, (2) devise effective algorithms for a multi-sensor version with different accuracy rates and ranges at different costs and with respective utilization limits in a partially-observed Markov decision process (POMDP) framework, (3) study multi-agent versions in MDP and POMDP frameworks, and (4) investigate adaptations of SDL variants and algorithms to other domains such as database query scheduling and stochastic project selection with dynamic learning. The underlying research problem has practical applications in important path planning environments such as robot navigation in stochastic domains, minefield countermeasures, adaptive traffic routing; and thus have a strong interdisciplinary aspect.

Semiconductors—-Devices and Sensors for Future-MEMS, Semiconductor Packaging and Test

Semiconductors are allowing society with ever more connectivity while the form factor keeps shrinking with introduction of hand-held mobile communications devices. So-called ‘Moore’s law’ predicts doubling of chip density every 18 months have been mostly realized for 4 decades and is expected to continue to be true. Chip designs are being released shorter time and wafer probing technologies face complex electrical and mechanical challenges as listed by ITRS 2014 Report (International Technology Roadmap for Semiconductors, www.itrs.net). Smart sensors, efficient new lighting and flexible display applications, OLED and LEDs are important promising new technologies based on semiconductors.

Clean Energy, Thin Films and Nanotechnology Based Nanocomposites enable new class of applications of bulk materials. More compact, highly sophisticated integration of insulators or conductors and devices are possible. Thin films of ferroelectrics or photovoltaic materials (solar cells) or specialty coatings lead to sustainable energy or building technologies and products. Thin film sensors are another category ranging from optical, chemical to ultrasonic devices. There are various thin film deposition technologies suitable to specific application from electroforming, physical vapor deposition, sol-gel methods. Lithography is required for pattern definition for some fine pitch custom applications. Conformal coatings are a different category of films used in the semiconductor or medical industry. Wind and solar energy are leading the charge in the new league for near-term global renewable energy.

Energy storage is the key to the future design of smart cities: Fuel cells and Li-ion battery systems. EVs, electric cars, battery driven cars will replace the combustion engines in future vehicles.

Fuel cells and approaching hydrogen economy

Solar cells for space applications and novel applications of CSP

Materials Selection for Optimal Design, High Performance Ceramics and Metal Alloys and Advanced Manufacturing enable strong wear-resistant tools or engine parts and in some cases medical implants. Materials such as silicon nitride, zirconia, Carbon composites are being used as well as specialty inert metals and other alloys. Methods of materials production and customization vary from synthesizing powders, sintering prototypes to plating components to laser micromachining and laser welding to final product. 3D printing revolution is changing the manufacturing and product development dramatically, with quicker design and proof of concept.

Biomaterials and biomechanics are gaining importance as implants and smart devices improve the quality of human life.

people-planet-profit.jpgTriple Bottom Line Sustainability Accounting

The phrase “the triple bottom line” was first coined in 1994 by John Elkington, the founder of a British consultancy called SustainAbility. His argument was that companies should be preparing three different (and quite separate) bottom lines. One is the traditional measure of corporate profit—the “bottom line” of the profit and loss account. The second is the bottom line of a company’s “people account”—a measure in some shape or form of how socially responsible an organisation has been throughout its operations. The third is the bottom line of the company’s “planet” account—a measure of how environmentally responsible it has been. The triple bottom line (TBL) thus consists of three Ps: profit, people and planet. It aims to measure the financial, social and environmental performance of the corporation over a period of time.

With the increasing concerns related to integration of social and economic dimensions of the sustainability into Life Cycle Assessment (LCA), a traditional LCA approach has been transformed into a new concept, which is called as “Life Cycle Sustainability Assessment (LCSA)”. However, when a conventional LCA is complemented by the economic input-output analysis, it is possible to capture all direct and indirect impacts considering the entire supply chain. Hence, I have mainly focused on the TBL-based sustainability accounting and developed economic input-output based holistic sustainability assessment model which integrates the triple bottom line sustainability assessment framework with supply chain management. This tool has been extensively used to analyze a wide range of policy issues related to environmental, economic, and social sustainability.

Current TBL-EIO model includes several macro-level sustainability indicators:

  1. Environmental Indicators: Energy, Water, Waste and Ecological Land Footprint
  2. Financial Indicators: Gross Domestic Product, Import, Profit, Total Economic Output
  3. Social Indicators: Employment, Income, Government Revenue, Health Expenditure, Work-related Injuries, Cost of externalities such as pollution, safety, etc

 

This holistic TBL-EIO based sustainability assessment tool has been applied for several multidisciplinary research areas, presented as follows:

  • Energy and the Environment
  • Sustainable Supply Chain Management
  • Sustainable Civil Infrastructure Systems
  • Green Manufacturing and Industrial Ecology

My research aims to develop decision support systems to analyze the social, environmental, and economic aspects of energy, manufacturing, supply chain and infrastructure systems from a life cycle perspective. I have been conducting research utilizing mainly the following decision making approaches:

  • Multi-Criteria Optimization
  • Monte-Carlo Simulation and Uncertainity Analysis
  • Linear Programming and Data Envelopment Analysis
  • Fuzzy-based Decision Making
  • Neural Networks

For further details about the TBL-EIO model, please see the following methodology papers and Argonne presentation:

Kucukvar, M., and Tatari, O. (2013). “Towards a triple bottom-line sustainability assessment of the U.S. construction industry“. The International Journal Life Cycle Assessment, Volume 18, Issue 5, pp 958-972.

Kucukvar, M., Noori, M., Egilmez, G., & Tatari, O. (2014). “Stochastic decision modeling for sustainable pavement designs”. The International Journal of Life Cycle Assessment, 1-15.

Onat, C., Kucukvar, M., and Tatari, O. (2014). “Integrating triple bottom line input-output analysis into life-cycle sustainability assessment: The case for U.S buildings .” The International Journal of Life Cycle Assessment, 1-18. 

Kucukvar, M. (2013). “Triple bottom line life cycle sustainability assessment framework for onshore and offshore wind turbines.” U.S. Department of Energy Office of Science, Argonne National Laboratory, IL, USA: Argonne Seminar

Please click for more detailed information.

  1. One-Dimensional Wide-Bandgap Semiconductor Nanostructures: Analysis and Applications

Research and development in nanotechnology has seen an astonishing progress during the past decade, and has now provided clearer indication of its potential. There is great potential to incorporate nanotechnology-enabled products and services into almost all industrial sectors and medical fields by 2020. Resulting benefits will include increased productivity, more sustainable development, and new jobs. We investigate silicon carbide (SiC), gallium nitride (GaN), and aluminum nitride (AlN) wide-bandgap semiconductor nanostructures and develop new applications of these nanostructures for energy generation, optoelectronics, and sensing. Our goal is not only to address the fundamental technical and scientific issues of 1D nanostructures, but also develop original products utilizing these materials. 1D-GaN and AlN nanostructures are extremely important for many applications in several fields including power transistors, optoelectronics, heat sinks, resonators, sensors, and nanogenerators. Similarly, due to its inherent superior properties, SiC is an excellent material for applications in many areas including microelectronics (high temperature, high power, and high frequency), thermoelectrics, optoelectronics, and biomedical.

Nanowire field effect transistor (NWFET) device.

GaN Nanowires

  1. Wide bandgap semiconductor devices for electric vehicles and grid applications

Innovation in power electronic components based on materials such as silicon carbide (SiC) and gallium nitride (GaN) can provide great solutions for a smaller footprint, higher current density and superior thermal management in various industry segments — Buildings and Industrial, Electronics and IT, Renewables and Grid Storage, and Transportation. Some specific examples are photovoltaic (PV) inverters, motor drives, and power converters for electric vehicles (EVs). These represent the fastest growing segment of the power electronics industry for the near future.

Application areas of power devices.