Home » Posts tagged 'Operations Research' (Page 2)

Tag Archives: Operations Research

Journal Paper accepted at IEEE Transactions on Industrial Informatics – An Axiomatic Design of a Multi-Agent Reconfigurable Mechatronic System Architecture

The LIINES is pleased to announce the acceptance of the paper “An Axiomatic Design of a Multi-Agent Reconfigurable Mechatronic System Architecture” to the IEEE Transactions on Industrial Informatics. The paper is authored by Prof. Amro M. Farid and Prof. Luis Ribeiro.

Recent trends in manufacturing require production facilities to produce a wide variety of products with an increasingly shorter product lifecycle. These trends force production facilities to adjust and redesign production lines on a more regular basis.

Reconfigurable manufacturing systems are designed for rapid change in structure; in both hardware and software components to address the required changes in production capacity and functionality.

Qualitative methods have recently been successful in achieving reconfigurability through multi-agent systems (MAS). However, their implementation remains limited, as an unambiguous quantitative reference architecture for reconfigurability has not yet been developed.

A design methodology based on quantitative reconfigurability measurement would facilitate a logical, and seamless transition between the five stages of the MAS design methodology, as shown below.

DesMethodology

Previous work on the reconfigurability of automated manufacturing systems has shown that reconfigurability depends primarily on architectural decisions made in stages 1, 2, 3, and 5. Operational performance of the manufacturing system after the reconfiguration is also important, but is often overlooked by the existing literature. As a result, it’s not clear:

  1. The degree to which existing designs have achieved their intended level of reconfigurability.
  2. Which systems are quantitatively more reconfigurable.
  3. How these designs may overcome their inherent design limitations to achieve greater reconfigurability in subsequent design iterations.

In order to address the previously mentioned issues with existing design methodologies, this paper develops a multi-agent system reference architecture for reconfigurable manufacturing systems driven by a quantitative and formal design approach, directly in line with the above Figure.

The paper uses Axiomatic Design for Large Flexible Engineering Systems to support a well-conceptualized architecture, which is necessary for excellent production system performance. Additionally, Axiomatic Design highlights potential design flaws at an early conceptual stage. This results in the first formal and quantitative reference architecture based on rigorous mathematics.

About the Author

Wester C.H. Schoonenberg completed his B.Sc. in Systems Engineering and Policy Analysis Management at Delft University of Technology in 2014. After his bachelors’ degree, Wester started his graduate work for the LIINES at Masdar Institute, which he continues as a doctoral student at Thayer School of Engineering at Dartmouth College in 2015. Currently, Wester is working on the integrated operation of electrical grids and production systems with a special interest in Zero Carbon Emission Manufacturing Systems.

A full reference list of LIINES publications can be found here:
WhiteLogo2

Follow LIINES Blog on WordPress.com

Share this post:

The LIINES Commitment to Open-Information

 Good science is reproducible.   This means that it must be publicly available, its contributions transparently communicated, and its data accessible.  These are principles that drive the everyday work of every individual’s research at the LIINES.  We now wish to go further and make a commitment to Open-Information.
Beginning today, the LIINES will seek to leverage its website to make all of its research 100% reproducible by the public at large.   This includes:
  • Sharing all input datasets used to conduct the research for which no prior proprietary or security commitments have been made.
  • Producing scientific publications in such a way that scientific peers can accurately verify & validate the work.
  • Making the content of all conference, journal and book-chapter publications freely available in author preprint form.  (Note: Most publishers allow self-archiving and open-distribution of author preprints).
We believe that the LIINES’ research has broad applicability to academia, industry, government and the public at large.   However, traditional publication venues are often only subscribed by academic universities and a handful of well-funded industrial companies.   All-too-often the people that can benefit from this work, never get a chance to see it.   This slows down the work’s potential impact and is counter to the LIINES mission.   It is for these reasons, that the LIINES makes its Open-Information commitment.
While it is natural that making all of this information available will take some time, we will be sure to keep blogging to keep you up to date of new additions to the LIINES website.  For now, feel free to visit the LIINES Datasets Repository which includes both data from our publications as well as a collation of several relevant and openly available datasets.
WhiteLogo2
Share this post:

Journal Paper Accepted at Applied Energy Journal – Demand Side Management in a Day-Ahead Wholesale Market: A Comparison of Industrial & Social Welfare Approaches

The LIINES is pleased to announce the acceptance of the paper entitled: “Demand Side Management in a Day-Ahead Wholesale Market: A Comparison of Industrial & Social Welfare Approaches” to Applied Energy Journal for publication. The paper is authored by Bo Jiang, Prof. Amro M. Farid, and Prof. Kamal Youcef-Toumi.

The intermittent and unpredictable nature of renewable energy brings operational challenges to electrical grid reliability. The fast fluctuations in renewable energy generation require high ramping capability which must be met by dispatchable energy resources. In contrast, Demand Side management (DSM) with its ability to allow customers to adjust electricity consumption in response to market signals has been recognized as an efficient way to shape load profiles and mitigate the variable effects of renewable energy as well as to reduce system costs. However, the academic and industrial literature have taken divergent approaches to DSM implementation. While the popular approach among academia adopts a social welfare maximization formulation, defined as the net benefit from electricity consumption measured from zero, the industrial practice introduces an estimated baseline.   This baseline represents the counterfactual electricity consumption that would have occurred without DSM, and customers are compensated according to their load reduction from this predefined electricity consumption baseline.

DR

In response to the academic and industrial literature gap, our paper rigorously compares these two different approaches in a day-ahead wholesale market context. We developed models for the two methods using the same mathematical formalism and compared them analytically as well as in a test case using RTS-1996 reliability testing system. The comparison of the two models showed that a proper reconciliation of the two models might make them dispatch in fundamentally the same way, but only under very specific conditions that are rarely met in practice. While the social welfare model uses a stochastic net load composed of two terms, the industrial DSM model uses a stochastic net load composed of three terms including the additional baseline term. While very much discouraged, customers have an implicit incentive to surreptitiously inflate the administrative baseline in order to receive greater financial compensation. An artificially inflated baseline is shown to result in a higher resource dispatch and higher system costs.

The high resource scheduling due to inflated baseline likely require more control activity in subsequent layers of enterprise control including security constrained economic dispatch and regulation service layer. Future work will continue to explore the technical and economic effects of erroneous industrial baseline.

About the Author:

Bo Jiang conducted this research in collaboration with her Master’s thesis advisor Prof. Amro M. Farid and Prof. Kamal Youcef-Toumi at Massachusetts Institute of Technology. Her research interests include renewable energy integration, power system operations and optimization. Bo is now pursuing her PhD at MIT Mechanical Engineering Department.

A full reference list of Smart Power Grids and Intelligent Energy Systems research at LIINES can be found on the LIINES publication page: http://engineering.dartmouth.edu/liines

 

WhiteLogo2

 

Share this post:

Journal Paper Accepted at Springer’s Intelligent Industrial Systems Journal: Multi-Agent System Design Principles for Resilient Coordination & Control of Future Power Systems

The LIINES is pleased to announce the acceptance of the paper: “Multi-Agent System Design Principles for Resilient Coordination & Control of Future Power Systems” in Springer’s Intelligent Industrial Systems Journal. The paper is authored by Amro M. Farid and was published online at May 28th 2015.

Recently, the vision of academia and industry has converged, defining future power system as intelligent, responsive, dynamic, adaptive, and flexible. This vision emphasizes the importance of resilience as a “smart grid” property. It’s implementation remains as a cyber-physical grand challenge.

Power grid resilience allows healthy regions to continue normal operation while disrupted or perturbed regions bring themselves back to normal operation. Previous literature has sought to achieve resilience with microgrids capable of islanded operation enabled by distributed renewable energy resources. These two factors require a holistic approach to managing a power system’s complex dynamics. In our recent work (e.g. link 1 and link 2), we have proposed as means of integrating a power system’s multiple layers of control into a single hierarchical control structure.

In addition to enterprise control, it is important to recognize that resilience requires controllers to be available even if parts of the power grid are disrupted. Therefore, distributed control systems, and more specifically Multi-Agent Systems have often been proposed as the key technology for implementing resilient control systems. Multi-agent systems are commonly used to distribute a specific decision-making algorithm such as those in market negotiation and stability control. However, very few have sought to apply multi-agent systems to achieve a resilient power system.

The purpose of the paper entitled “Multi-Agent System Design Principles for Resilient Coordination & Control of Future Power Systems” is two fold. First, it seeks to identify a set of Multi-Agent System design principles for resilient coordination and control. Second, the paper assesses the adherence of existing Multi-Agent System implementations in the literature with respect to those design principles.

The set of design principles is based on newly developed resilience measures for Large Flexible Engineering Systems. These measures use Axiomatic Design and are directly applicable to the power grid’s many types of functions and its changing structure. These design principles, when followed, guide the conception of a multi-agent system architecture to achieve greater resilience.

About the author: Wester C.H. Schoonenberg completed his B.Sc. in Systems Engineering and Policy Analysis Management at Delft University of Technology in 2014. After his bachelors’ degree, Wester started his graduate work for the LIINES at Masdar Institute, which he continues as a doctoral student at Thayer School of Engineering at Dartmouth College in 2015. Currently, Wester is working on the integrated operation of electrical grids and production systems with a special interest in Zero Carbon Emission Manufacturing Systems.

Screen Shot 2015-02-07 at 11.38.55

LIINES Website: http://amfarid.scripts.mit.edu

Share this post:

Journal Paper Accepted at the Journal of Intelligent Manufacturing: Measures of reconfigurability and its key characteristics in intelligent manufacturing systems

The LIINES is pleased to announce that the Journal of Intelligent Manufacturing has accepted our paper entitled: “Measures of reconfigurability and its key characteristics in intelligent manufacturing systems”. The paper is authored by Amro M. Farid and was published in October 2014.

Many manufacturing challenges arise with the global trend of increased competition in the marketplace.  Production processes must deal with shorter product lifecycles and mass-customization. Consequently, production systems need to be quickly and incrementally adjusted to meet the ever-changing products. Reconfigurable manufacturing systems have been proposed as a solution that facilitates changing production processes for highly automated production facilities.

Much research has been done in the field of reconfigurable manufacturing systems. Topics include: modular machine tools and material handlers, distributed automation, artificially intelligent paradigms, and holonic manufacturing systems.  While these technological advances have demonstrated robust operation and been qualitatively successful in achieving reconfigurability, there has been comparatively little attention devoted to quantitative design methodologies of these reconfigurable manufacturing systems and their ultimate industrial adoption remains limited.

Measuring reconfigurability of manufacturing systems quantitatively has been a major challenge in the past, since a quantitative reconfigurability measurement process was non-existent. Earlier work developed a measurement method that extracts measurables from the production shop floor. When this was established, basic measures of reconfiguration potential and reconfiguration ease were developed, based on axiomatic design for large flexible engineering systems and the design structure matrix respectively.

Reconfiguration of a production process can be split up in four steps: Decide which configuration, Decouple, Reorganize, and Recouple. The larger the number of elements in the system, the more configurations are made possible. This is measured using the reconfiguration potential measure, based on axiomatic design for large flexible engineering systems.

Production processes contain multiple interfaces within themselves. Multiple layers of control can be distinguished, that have to work together to coordinate the physical components. These interfaces are the main determinants for the reconfiguration ease measure.

This paper combines these techniques to define a quantitative measure for reconfigurability and its key characteristics of integrability, convertibility and customization.    The intention behind this research contribution is that it may be integrated in the future into quantitative design methodologies for reconfigurable manufacturing systems, which may be easily adopted by industrial automation and production companies.

About the author: Wester Schoonenberg completed his B.Sc. in Systems Engineering and Policy Analysis Management at Delft University of Technology in 2014. After his bachelors’ degree, Wester started his M.Sc. at Masdar Institute of Science & Technology. Currently, Wester is working on the integrated operation of electrical grids and production systems with a special interest in the demand side management of industrial facilities.

Screen Shot 2015-02-07 at 11.38.55

LIINES Website: http://amfarid.scripts.mit.edu

Share this post:

Journal Paper Accepted at IEEE Transactions on Industrial Electronics: An Enterprise Control Assessment Method for Variable Energy Resource Induced Power System Imbalances. Part 2: : Parametric Sensitivity Analysis

We are happy to announce that our recent paper entitled: “An Enterprise Control Assessment Method for Variable Energy Resource Induced Power System Imbalances. Part 2: Parametric Sensitivity Analysis”, has been accepted to IEEE Transaction on Industrial Electronics. The paper is authored by Aramazd Muzhikyan, Prof. Amro M. Farid and Prof. Youcef Kamal-Toumi.

The variable and uncertain nature of the variable energy resources (VER) introduces new challenges to the balancing operations, contributing to the power system imbalances. To assess the impact of VER integration on power system operations, similar statistical methods have been used by renewable energy integration studies. The calculations are based on either the net load variability or the forecast error, and use the experience of power system operations. However, variability and forecast error are two distinguishing factors of VER and both should be taken into consideration when making assessments.

This paper uses the methodology from the prequel to systematically study the VER impact on power system load following, ramping and regulation reserve requirements. While often ignored, the available ramping reserve reflects the generation flexibility and is particularly important in the presence of VER variability. This provides a detailed insight into the mechanisms by which the need for additional reserves emerges. The concept of enterprise control allows studying the impact of power system temporal parameters as well as net load variability and forecast error holistically.

The application of an enterprise control assessment framework allows the empirical identification of the most influential parameters different types of resource requirements. The inclusion of the power system temporal parameters, such as day-ahead market (SCUC) and real-time market (SCED) time steps, is a particularly distinguishing feature of the work. Use of the case-independent methodology allows generalization of the results and prediction of how the system resource requirements change when one of the parameters varies. Moreover, the results reveal the degree of importance of each lever for the power system reliable operations which is crucial for the strategic planning of the grid modernization.

errDAImpactResSTD scedImpactRegSTD

Share this post:

Duke Energy on Analytics and the Internet of Things

It’s been a long time since 2003 when the concept of the Internet of Things was first proposed by U. of Cambridge Auto-ID Laboratory.  At the time, Dr. Amro M. Farid, now head of the Laboratory for Intelligent Integrated Networks of Engineering Systems, was a doctoral student investigating how RFID technology enabled intelligent products within reconfigurable manufacturing systems.  The Internet of Things was being applied primarily in the manufacturing and supply chain domain.

Since then, the Internet of Things concept has taken hold not just in manufacturing systems and supply chains but nearly every industrial system domain including energy.    Every “thing” or “device” has the potential to be connected via an intelligent sensor so as to make decisions — be they centralized within an operations control center — or distributed amongst artificially intelligent multi-agent systems.   The Internet of Things concept has the potential to fundamentally transform industrial systems.

Have a look at Duke Energy’s take on the Internet of Things:

The LIINES is proud to have been working in this area since its inception and continue to do so.  More information on our research can be found on the LIINES website.

WhiteLogo2

LIINES Websitehttp://amfarid.scripts.mit.edu

Share this post:

Journal Paper Accepted at ISA Transaction: Event Triggered State Estimation Techniques for Power Systems with Integrated Variable Energy Resources

The LIINES is happy to announce that ISA Transactions has accepted our recent paper entitled: Event Triggered State Estimation Techniques for Power Systems with Integrated Variable Energy Resources.  The paper is authored by Reshma C. Francy, Prof. Amro M. Farid and Prof. Kamal Youcef-Toumi.
In recent years, we have had the opportunity to contribute to two large studies that present visions of the future smart grid:  The MIT Future of the Electric Grid Study, and the IEEE Vision for Smart Grid Controls: 2030 and Beyond.  Both of these works emphasized that in order for the future grid to be truly smart, it has to be responsive, dynamic, adaptive and flexible.  This is the case even when highly variable renewable energy sources sources are plugged in.   The first step in achieving this vision is having greater “situational awareness” — knowing what is going on when and where in the grid.
OCC
For decades, state estimation has been a critical technology in achieving such situational awareness for power system operators.   Over time, it has become quite the mature technology. But, the integration of renewable energy changes all that.  Not only does it introduce rapidly changing behavior into the grid; but it also does so in the low voltage distribution system where state estimation is not usually applied.   The conventional solution is to not just monitor the grid faster but also for the entire power grid all the way down to the low voltages.  That means that not only do all the power grid’s measurements have to be gathered from across power grid’s geography but they also have to computed at an ever faster rate.   This is an exponentially growing problem  — hardly a solution befitting a future “smart” grid.
This paper seeks to address these two requirements in a practical way.   The idea is to use a concept called “event-triggering”.  It takes advantage of the fact that the wind doesn’t always blow and the sun doesn’t always shine.  When local power grid conditions are highly variable, say at a wind turbine or solar panel, a “trigger” will kick in telling the state estimator to run.  But when the power grid is relatively stable, the new state estimator will use a simplified linear approach based upon the last time the full state estimator was run.  Relative to traditional state estimation, this simple solution has been shown to reduce computational time by 90% in numerical case studies.
While ultimately, in the long term, the smart grid will require a fundamental “rethink” in how to approach state estimation, monitoring, and situational awareness, this solution demonstrates how traditional state estimation techniques can be enhanced for future smart grid applications.
A full reference list of smart grid research at LIINES can be found on the LIINES publication page: http://amfarid.scripts.mit.edu

WhiteLogo2

LIINES Website: http://amfarid.scripts.mit.edu

Share this post:

Prof. Amro M. Farid gives invited lecture at MIT Transportation Seminar Series

On December 5, 2014, Prof. Amro M. Farid gave an invited lecture at the MIT Transportation Seminar Series (Cambridge, MA, USA).   The presentation entitled:  “Intelligent Transportation-Energy Systems for Future Large Scale Deployment of Electrified Transportation” featured the LIINES’ latest research in transportation electrification.

The presentation advocates an integrated approach to transportation and energy management.  At its core, the intelligent transportation energy system (ITES) requires a new transportation electrification assessment methodology that draws upon microscopic traffic simulation, power grid dynamics, and Big Data-Driven use case modeling. Such an ITES would come to include coupled operations management decisions including: vehicle dispatching, vehicle routing, charging queue management, coordinated charging, and vehicle-to-grid ancillary services.  The presentation also featured the results from the first full scale electric vehicle integration study which was recently conducted for a taxi-fleet use case in Abu Dhabi.   The study suggests a close collaboration between the Abu Dhabi Department of Transportation and the Abu Dhabi Water and Electricity Authority in future large scale deployments of electrified transportation.

The presentation draws heavily from several LIINES publications including the UAE State of Energy Report, the UAE State of the Green Economy Report, the first hybrid dynamic model for transportation electrification.  The results of this first full-scale study were first presented publicly at the 2nd IEEE International Conference on Connected Vehicles & Expo held December 2-6, 2013 in Las Vegas, NV, USA, and the Gulf Traffic Conference held December 9-10 2013 in Dubai, UAE.  These presentations demonstrated a successful collaborative project between Masdar Institute, the Abu Dhabi Department of Transportation, and Mitsubishi Heavy Industries.

In depth materials on LIINES research on transportation electrification can be found on the LIINES publication page:  http://amfarid.scripts.mit.edu

WhiteLogo2

LIINES Website: http://amfarid.scripts.mit.edu

Share this post:

Journal Paper Accepted at Applied Energy Journal: Quantitative engineering systems modeling and analysis of the energy-water nexus

The LIINES is happy to announce that Applied Energy Journal has accepted our recent paper entitled:  “Quantitative engineering systems modeling and analysis of the energy–water nexus” for publication.  The paper is authored by William N. Lubega and Prof. Amro M. Farid.  

Electric power is required to extract, condition, convey, dispose of and recycle water for human use. At the same time, the bulk of global electricity generation capacity uses water as a heat sink or prime mover. This energy-water nexus is of growing importance due to increased demand for water and electricity; distortion of the temporal and spatial availability of fresh water due to climate change; as well as various drivers of more energy-intense water supply for example increased wastewater treatment requirements, and more water-intense electricity generation for example emissions control technologies at power plants.

water-energy_lg

There are several notable published studies on this nexus. At a technology level, there have been attempts to optimize coupling points between the electricity and water systems to reduce the water-intensity of technologies in the former and the energy-intensity of technologies in the latter. Empirical determinations of the electricity-intensity of water technologies and the water-intensity of electricity technologies have been reported and analyzed. Various models that enable the exploration of the water resource implications of defined electricity sector development pathways and thus support the analysis of various water and electricity policies have also been developed. To our knowledge however, a transparent physics-based approach that interfaces a model of the electricity system to models of the municipal water and wastewater systems enabling an input-output analysis of these three systems in unison has not been presented. Such a modeling approach would support integrated control applications as well as integrated planning without a priori specification of development pathways, for example through optimization.

A paper recently published by the LIINES in Applied Energy titled Quantitative engineering systems modeling and analysis of the energy–water nexus presents such a systems-of-system model. In this work, bond graphs are used to develop models that characterize the salient transmissions of matter and energy in and between the electricity, water and wastewater systems as identified in the reference architecture. Bond graphs, which are graphical representations of physical dynamic systems, were chosen as the modeling tool as they facilitate the inter-energy-domain modeling necessitated by the heterogeneous nature of the energy-water nexus. Furthermore they clearly identify causality and readily allow for model enhancement as required by applications. The developed models, when combined, make it possible to relate a region’s energy and municipal water consumption to the required water withdrawals in an input-output model.  This paper builds on another LIINES publication entitled “A Reference Architecture for the Energy-Water Nexus” found in the IEEE Systems Journal.

lubeg1

 

This research is of particular significance to countries in the Gulf Cooperation Council, all of which have limited fresh water resources and thus depend on energy-expensive desalination to meet a large portion of their water needs. This dependence enhances the degree of coupling between the electricity and water systems and thus the associated vulnerability concerns. Furthermore, motivated by the cogeneration of electric power and desalinated water, combined electricity and water authorities have been established in the region. The multi-energy domain model developed in this work is therefore of immediate relevance to the planning and control efforts of these existing institutions.

 

About the Author:

William N. Lubega conducted this research in collaboration with his Master’s thesis advisor Prof. Amro M. Farid in LIINES at the Masdar Institute of Science & Technology Engineering Systems & Management Department.  William is now a doctoral research assistant at the University of Illinois Urbana-Champaign Civil & Environmental Engineering department as part of the Energy-Water-Environment Sustainability Track.  There, he continues his energy-water nexus research in the Stillwell Research Group.

A full reference list of energy-water nexus research at LIINES can be found on the LIINES publication page: http://amfarid.scripts.mit.edu

WhiteLogo2

LIINES Website: http://amfarid.scripts.mit.edu

Share this post:

Subscribe

Enter your email address to receive notifications of new posts by email.

Join 585 other subscribers