Browsing by Author "Rakai, Logan"
Now showing 1 - 8 of 8
Results Per Page
Sort Options
- ItemOpen AccessA Multilevel Congestion-Based Global Router(2009-11-17) Rakai, Logan; Behjat, Laleh; Areibi, Shawki; Terlaky, TamasRouting in nanometer nodes creates an elevated level of importance for low-congestion routing. At the same time, advances in mathematical programming have increased the power to solve complex problems, such as the routing problem. Hence, new routing methods need to be developed that can combine advanced mathematical programming and modeling techniques to provide low-congestion solutions. In this paper, a hierarchical mathematical programming-based global routing technique that considers congestion is proposed. The main contributions presented in this paper include (i) implementation of congestion estimation based on actual routing solutions versus purely probabilistic techniques, (ii) development of a congestion-based hierarchy for solving the global routing problem, and (iii) generation of a robust framework for solving the routing problem using mathematical programming techniques. Experimental results illustrate that the proposed global router is capable of reducing congestion and overflow by as much as 36% compared to the state-of-the-art mathematical programming models.
- ItemOpen AccessData Analytics in Competitive Electricity Markets to Uncover the Impact of Emerging Technologies(2017) Zamanidehkordi, Payam; Zareipour, Hamid; Rakai, Logan; Karki, Rajesh; Far, Behrouz; Knight, Andrew; Nowicki, Edwin; Hollis, AidanThe electrical power industry has entered a transition towards sustainable, reliable and clean solutions. It is a continuous revolution trending to a large-scale expansion of renewables in power systems. There have been, however, serious concerns over reliable and secure operation of power systems. Energy storage facilities are increasingly being used to help integrate renewable energy resources into the grid. While understanding the environmental benefits of these emerging technologies is straightforward, the economic impacts of their integration in a competitive market is more complicated. These emerging technologies are likely to have an economically-important effect on the dynamics of electricity prices. This is a concern to different sections of electricity markets including power suppliers, policy makers, and end users. This thesis focuses on applying data mining tools to competitive electricity markets in order to uncover the impact of emerging technologies such as wind power and storage systems on the dynamics of electricity prices. Data-driven approaches are developed to explore the impact on wholesale prices of individual wind farms and independently-operated large-scale energy storage systems. Additionally, this thesis proposes a data-driven methodology to determine a justified support scheme for upcoming wind farms by incorporating their estimated revenue and levelized cost of energy. Moreover, an operation-inspired electricity price prediction scheme is developed to improve the economic profit obtained from operation of storage facilities in competitive markets. Numerical simulations are provided for the electricity markets of Alberta and Ontario. The results prove the efficiency and accuracy of proposed methodologies in estimating the impact on wholesale prices of emerging technologies. In addition, the obtained results from both competitive markets indicate that the presented methodology in this thesis is able to estimate the revenue of an upcoming wind farm with reasonable accuracy, which successively determines the support scheme awarded to the project. Moreover, the performed analyses manifest the effectiveness of the proposed price prediction scheme in improving the economic performance of storage systems.
- ItemOpen AccessFast and effective methods for alleviating congestion in ilp-based global routing(2008) Rakai, Logan; Behjat, Laleh; Dimitrov, Vassil
- ItemOpen AccessMethods for Solving Modern, Scale-Borne Problems in VLSI Physical Design(2012-10-01) Rakai, Logan; Behjat, LalehThe design automation community is confronted with new challenges every technology node. Many of the challenges are borne out of issues relating to scale, be it the very small or the very large. For example, the extremely large scale of the number of instances in modern designs creates challenges in effectively exploring vast solution spaces in reasonable amounts of time. At the other end of the scale spectrum, the extremely small scale of features created by modern lithography processes are highly susceptible to process variations which affect performance and yield. This thesis deals with the development of methods for solving scale-borne problems in the physical design of integrated circuits. This thesis addresses challenges faced in two important phases of physical design: placement and clock network synthesis. The importance of these two phases is reflected in the fact that they have been the subject of five out of the last seven ACM/IEEE International Symposium on Physical Design (ISPD) contests. The number of instances and the size of the solution space in performing placement are truly immense. A proven linear-time clustering algorithm is proposed to deal with the explosion of problem sizes being encountered. Several extensions to the algorithm are proposed to further improve the quality of results. The number of instances in clock network design is also growing at a rapid pace. In order to cope with this challenge, a generic framework to parallelize algorithms that perform the main stages of clock network synthesis is proposed. Theorems are provided to prove asymptotically optimal speedup when the framework is applied to several classes of algorithms. Another challenge addressed regarding clock network synthesis is that of variation. A method is proposed for handling variations in lengths and widths of buffers and interconnects, that arise from the manufacturing process.
- ItemOpen AccessA Multilevel Congestion-Based Global Router(Hindawi Publishing Corporation, 2009-09-02) Rakai, Logan; Behjat, Laleh; Areibi, Shawki; Terlaky, Tamas
- ItemOpen AccessA New Length-Based Algebraic Multigrid Clustering Algorithm(Hindawi Publishing Corporation, 2012-02-25) Rakai, Logan; Farshidi, Amin; Behjat, Laleh; Westwick, David
- ItemOpen AccessOn Time Aggregation Techniques for Power System Planning Applications(2023-01-10) Sarajpoor, Nima; Zareipour, Hamidreza; Rakai, Logan; Messier, Geoffrey; Leung, Henry; Wood, DavidDue to the evergrowing renewable energy resources and their penetration in the electrical grid, the power system operators face the difficulty of handling such uncertainty while obtaining the result of operation and planning studies in a tractable time. This thesis is focused on developing time aggregation frameworks for compressing renewable energy as well as electricity demand data into a limited number of representative periods from which the complex power system studies can be solved in a reasonable time without reducing the loss of accuracy. First, an overview of different parts of a clustering process is provided. The pros and cons of methods are discussed to help the reader understand the rationale behind the author narrows down the comparative methods in this thesis. A section is also provided to discuss the centroid selection process, which is often an optional step in the clustering process. Yet, it is a crucial step in the time aggregation in the context of power system studies. This thesis tackles the time aggregation challenge by developing a method based on an elastic-based distance that can reflect the volatility of time series data such as wind power while preserving its co-movement with electricity demand. Next, the spatiotemporal factor is taken into account and a method is proposed to handle time aggregation in the presence of several renewable energy resources. Finally, this thesis brings in the concept of stability, and develops a framework that can help with obtaining a set of representative periods that are more stable compared to the existing methods. In all of the aforementioned works, the proposed approach is compared against a set of comparative methods considering both data- and model-based evaluation. Regarding the former, the methods are compared according to their performance in reflecting different characteristics of data. In the latter, the method is evaluated in the context of a power system problem in an electrical network. To better show the impact of our method in preserving shape, energy storage units are added to the network. Different indices are measured to reflect the performance of the proposed approach from different perspectives.
- ItemOpen AccessPower and Timing Driven Optimal Gate, Clock Buffer and Clock Wire Sizing in High Performance Digital Integrated Circuits(2016) Farshidi, Amin; Behjat, Laleh; Westwick, David; Dimitrov, Vassil; Rakai, Logan; Yanushkevich, Svetlana; Anjos, MiguelGate sizing and clock buffer and wire sizing are intertwined problems that greatly impact the trade-off between the power consumption and timing metrics of digital integrated circuits. With the increasing demands for mobile devices and low power technologies, the power consumption has become as important as the timing performance for the integrated circuit designers. However, finding a balanced trade-off among these objectives is a complex task that may need time-consuming experiments. On the other hand, in the recent technology nodes, the effects of process variations in the circuit component sizes cannot be neglected. In this thesis, a circuit optimization framework is proposed to handle the competing objectives and solve the multi-objective geometric programming problem by achieving a balanced trade-off between power and timing metrics. The proposed framework is self-tuning meaning that the multi-objective weights are optimally calculated during the optimization procedure without any manual tuning by the designer. In the next stage, robust optimization is employed to develop the robust geometric programming counterpart of the uncertainty-aware self-tuning multi-objective optimization framework. It is proposed to consider the buffer size variations during the optimization process by incorporating an uncertainty model in a robust optimization framework. Then, a smart heuristic for discretization of the solutions of the proposed frameworks is developed that remedies the timing performance degradations due to the discretization. Finally, a guideline is provided for the designers to decide which one of the proposed clock network buffer sizing frameworks is the most appropriate for their design goals.