Student Experiences


Jump to Summer 2013
Jump to Summer 2014
Jump to Winter 2015
Jump to Summer 2015
Jump to Summer 2016
Jump to Summer 2017
Jump to Summer 2018

Summer 2018

Project Title:  Validation of WindSim CFD Software in Complex Terrain 

Participants:  Daniel Tabas (JHU), Jiannong Fang (EPFL) 

Objectives: Using results from the summer project, begin to write a journal paper and conduct new simulations

Activities: Discussing new simulation results and writing

Conclusions: We found that the actuator disc model and forest model both play important roles in the accuracy of the simulations. In addition, different parametrizations for the actuator disc model provide different levels of accuracy.


Project Title: Synthetic Network Models of Electric Power Grids

Participants: Colin Bowen, Rafa Espejo González, Dra. Sara Lumbreras Sancho, Dr. Andrés Ramos Galan

Objectives: To improve a network model of the electric power grid using geographical and electrical properties as parameters.

Activities: Creating an algorithm to evaluate the viability of transmission line based on actual geographical obstacles that would present challenges in the cost and construction of a transmission line.

Conclusions: Improvements and features are being added to the algorithm. Following this, other improvements will be added to the model. The improved model will then be the basis for a paper to be submitted for publication.


Project Title: Strategic Generation and Transmission Expansion Planning Under Uncertainty in the Electricity Market — Case Study

Participants: Charles Gulian, Diego Tejada, Isaac Gonzalez, Sonja Wogrin

Objectives: Our objective was to test and compare different techniques for optimizing investments in new transmission and generation facilities (including wind power and renewable generators) under uncertainty in future available wind power.

Activities: We compared four optimization-under-uncertainty techniques (Deterministic Expected Value Programming, Robust Programming with Mini-Max Cost, Robust Programming with Mini-Max Regret, and Stochastic Programming) in a model power grid (simulated with the Pyomo Python package) with 5 existing nodes with power demand and 4 potential investment decisions in new wind power OR thermal generation and up to 20 potential investment decisions in new transmission lines between nodes and new generators. Our power grid model sought to minimize the total cost of generation/transmission investments and system operational costs. We compared the planning solutions of the various optimization techniques under a variety of wind power scenarios and power grid conditions.

Conclusions: We found that Deterministic Expected Value Programming–the simplest and most naive of the four optimization techniques that we tested–generally performed poorly, as it represents future hourly wind power as the average of historical data, and therefore overestimated the proportion of time that there was available wind power. Therefore, Deterministic Expected Value Programming invested in more wind power than other techniques. On the other hand, Robust Programming with Mini-Max Cost invested in the least new wind power and the most new thermal generation because it optimizes for the “worst-case scenario”– i.e. the scenario with the least future wind power available to the system. Stochastic Programming and Robust Programming with Mini-Max Regret made similar investment decisions in most scenarios, and tended to make the best planning decisions in terms of future system costs under different future wind scenarios. More research is needed to determine the subtle differences between these techniques.

Project Title:  Physical and Chemical Characterization of Effluent Blackwater in the NEST Water Hub at the Swiss Federal Institute of Aquatic Science and Technology (Eawag)

Participants:  Thomas Howard

Objectives: Determine the quantities and qualities of incoming wastewater in the NEST WaterHub at the Swiss Federal Institute of Aquatic Science and Technology (Eawag). Determine why and how blackwater quantities and qualities (Q&Q) vary over time (hours and days) by characterizing the liquid and solid streams from the building’s Aquatron solid-liquids separator.

  1. Analyze the performance of the Water Hub’s Aquatron installment. Quantify the Aquatron’s solid-liquid separation efficiency and variability, and prescribe changes to optimize the Aquatron’s separation efficiency?

Activities:  As an intern at Eawag, I initiated experiments and developed the standard operating procedure for the Water Hub’s Aquatron solid-liquid separator. I modeled and quantified daily wastewater production trends and flow rates in this innovative working/living environment. Furthermore, this summer, I performed chemical and physical analyses to characterize various properties of the building’s effluent blackwater, including chemical oxygen demand, electric conductivity, total nitrogen, total solids, volatile solids, total suspended solids, volatile suspended solids, pH, and ammonium concentration. These measured quantities and qualities were used in assessing the efficiency of the building’s Auqatron installment, and they will be used in the design of bioreactors to study improved methods of blackwater dewatering. This summer, I also tested a number of polymer conditioners to facilitate blackwater dewatering and began experiments using a Bucher press to study the feasibility of converting wastewater into usable energy supplies.

Conclusions:  My studies this summer showed that the Aquatron installment in the NEST Water Hub has over an 80% solid-liquid separation efficiency by mass. On average, the Aquatron’s separated solids stream contained 7 times more dissolved solids and suspended solids than the liquids stream did. Moreover, the solids stream blackwater had a chemical oxygen demand 11 times higher than that of the liquids stream blackwater. The measured COD and TS values were very strongly related, and a 1 mg/L increase in the TS of a sample was a reliable predictor of an approximately 1500 mg/L increase in the COD of a sample. My work also showed that recycled wastewater could be effectively dewatered using polymer conditioners at low doses, and it will be used in further studies on the dewaterability of blackwater and feasibility of energy production and resource recovery from wastewater.

Special Thanks: Special thanks are owed to BJ Ward, Stanley Sam, and Linda Strande for guiding, supervising, and mentoring me through this project. I thoroughly enjoyed working at Eawag, and I sincerely appreciate the time and effort that you all put in to provide me with this incredible experience. This summer has had an indelible impact on me, and I am inspired by the work you all do. I am also in great debt to Ben Hobbs, Lynn Roberts, and Shahin Zand for making this opportunity possible. It is not without your support that I would have had this opportunity. Thank you for being mentors, advocates, advisers, and role models for me.


Summer 2017

 Project Title: Cost Allocation for Transmission Expansion Planning

Participants: Laura Supple – Ing. Deniz Sun, Dr. Michel Rivier Abbad, Dr. Luis Olmos Camacho

Objectives: This research investigates the effect of different cost allocation mechanisms on electricity transmission expansion planning into the near future (10-20 years). The purpose of this project was to review several methodologies for quantifying and monetizing the human health and ecological damages caused by fossil-fired electricity generation to determine an appropriate marginal cost of power generation compatible with the beneficiaries’ method of cost allocation.

Activities: Activities of this research project were focused on the review of several methodologies quantifying and monetizing human health and/or ecological damages caused by local air pollutants emitted during fossil-fired electricity generation, with a primary focus on the European Union. These methodologies included the Impact Pathways Approach developed by the ExternE project for the European Commission, appropriate tax estimates for fuel based on marginal external damages of consumption by the International Monetary Fund, The Uniform World Model for pollution dispersion and environmental damage developed at the Basque Center for Climate Change, and the Greenhouse Gas – Air Pollution Interactions and Synergies  (GAINS) model developed by the International Institute for Applied Systems Analysis. The review consisted of an assessment of methodological robustness and the accuracy of data or assumptions used, comprehensiveness of the damages assessed, applicability of the results to the beneficiaries’ method of cost allocation, replicability of the results, and overall strengths and weaknesses of each.

Conclusions: After thorough comparison of the four above methodologies it was decided that both the International Monetary Fund and GAINS estimates of damages from power generation were not suitable for this research. The IMF estimates were not specific enough to the European context and relied on highly simplified parametrizations of pollution dispersion that raised concerns over the accuracy of results. The GAINS model was deemed unsuitable due to the unclear connection between sources and receptors of pollution, which is necessary for the beneficiaries’ method of cost allocation. The Impact Pathways Approach from the ExternE program and the Uniform World Model were deemed the most suitable methods for estimating damages from power generation. In the next phases of research, project members will consider these two methods in the context of transmission expansion planning and attempt to connect decreases in local air pollution due to increased renewable generation to human health and ecological benefits.

Project Output: Summary presentation of findings from each methodology (see attached powerpoint).

Personal Experiences: Through this research I’ve been able to work with brilliant scientists, engineers, and economists from around the world, brought together by their pursuit of common solutions to large-scale challenges. Through the WINDINSPIRE program I’ve also been able to participate in two international wind energy conferences – the WINDFARMS 2017 Conference hosted in Madrid and the Wind Energy Science Conference hosted in Copenhagen– giving me the rare chance as an undergraduate to view topics of public policy, technology, and environmental science from the perspective of the researchers, where the unknown is just a new opportunity for discovery. My experience with WINDINSPIRE has given me invaluable exposure to the world of research and new perspectives to create an academically, socially, and culturally thrilling summer!

Title: Reducing Temporal Information in Representing Storage in Medium- and Long-Term Unit Commitment Models

Participants: Maya Domeshek (Smith College, Physics Undergraduate), International Hosts: Sonja Wogrin (Universidad de Comillas), Diego Alejandro Tejada Arango (Universidad de Comillas)

Objectives: Compare different methods of simplifying the representation of time-dependent information (such as variable demand and wind availability) in medium- and long-term power system models involving energy storage.

Activities: An hourly unit commitment model was compared to four models with approximate temporal information.  These models included a Representative Period model, which is a standard technique that performs all calculations on a representative subset of days from the year; a System States model, which was previously developed by Wogrin and performs calculations for a subset of hours in the year while preserving some chronology between them; a new system states model, which reduces the calculation time by being even more approximate in its treatment of storage than the old system states model; and a model that combines Representative Periods with the chronology preserving aspects of the System States model. My part of the project involved writing Matlab code to simplify the data for the models, experimenting with ways to guarantee that the energy storage remained within bounds, and executing the models.

Conclusions: While each approximate method has advantages and disadvantages, the model that combined a representative period approach with the chronological aspects of the system states was best able to represent both long- and short-term storage units.

Project Output: A paper is being prepared for publication

Personal Experiences: I was able to learn about field that was entirely new to me but so integral to the day-to-day functioning of the modern world and to experience university collaboration in a new context.

Participants: Danielle Jacobson, Farshid Kardan, Fernando Porte-Agel

Objective: The purpose of this study is to use computational fluid dynamics simulations to develop a wind flow map of the EPFL campus, chosen to represent a complex urban morphology. This wind map can be used to evaluate wind comfort and safety on campus as well as identify locations and predict potential power outputs for urban wind turbines.

Activities: A 3D model of the EPFL campus was created with the architecture software Rhinoceros, taking care to represent details. Next, a tetrahedral mesh was generated for two wind directions- north and south- chosen after analysis of meteorological data. Reynolds-Averaged Navier-Stokes (RANS) equations along with two different turbulence models were used for the solver- the Reynolds Stress Model (RSM) and a transitional flow model (SST Re-γθ). The inlet wind profile was set to atmospheric boundary layer conditions with the different surface types assigned varying roughness lengths, roughness constants, and sand grain roughness heights (z0, Cs, and ks,respectively). The simulations were run using the commercial CFD software ANSYS Fluent on the SCITAS cluster at EPFL.

Conclusions: The mesh was refined using edge sizing of 0.2 m to obtain an average skewness of 0.22 for both north and south wind directions consisting of nearly 24.5 million cells, indicating a very high quality mesh categorized as “excellent” according to skewness value (>0-2.5). Vertical cross sections will be taken in the north, east, south, and west directions surrounding the measurement tower, offset by 20 cm, in order to compare simulated results to measured data from the tower. A horizontal cross-section showing the wind comfort class at pedestrian height will also be produced. Vertical cross sections of areas of high velocity will be analysed for possible turbine placement.

Figure 1 Simulated wind velocity magnitude in m/s at pedestrian height (1.75 m)

Output: poster for WESC, contribution towards the paper “Evaluation of RANS turbulence models in the simulation of flow over an idealized elongated building using different grid resolutions and structures”, final presentation (pending)

Personal Experience: WINDINSPIRE allowed me to go places and have experiences I otherwise wouldn’t have been able to. Travelling to Copenhagen for the Wind Energy Science Conference was both an enriching academic and cultural experience, as I had never been to Scandinavia before then. I was also able to travel to Turkey and Italy in my spare time, as well as explore lots of Switzerland through some of the most beautiful hikes I have ever done. Lausanne was a beautiful and vibrant city to live in, and I will miss it very much!

Project Title: Long Term Case Study for TEPES using e-Highway 2050 Data

Participants: Cole Rosenberger, Dr. Sara Lumbreras Sancho, Dr. Andres Ramos Galan, Quentin Ploussard


Using the demand, generation, and network data from the e-Highway 2050 report, and the TEPES model, design a case study for each of the long-term scenarios.  Each case study follows the outline of the TEPES interface.  After each long-term case study is produced, TEPES will be executed using GAMS, with the result being a finalized TEP for each scenario.  The optimized TEP results from TEPES can be compared against each other, and against the TEP results from the e-Highway 2050 report to draw conclusions about the differences between the long-term scenarios, and about the assumptions made by e-Highway for their TEP results.

Activities: In order to gain a better understanding of the objective and the model, I began by reading several papers on the TEPES model.  Once I understood the optimization techniques used by TEPES, and the way that the Excel interface was read by GAMS, I moved on to reading the resulting reports published by the e-Highway 2050 project.  From there, I was able to download the generation and demand data for each scenario, and manipulated it in such a way that made it easier to input into the TEPES interface.  As I entered more data, the file size of the interface grew to be too large, so I decided to separate each long-term scenario into its own model, thus producing five models, one for each long-term scenario.  At the end of my summer period, I had produced five working models, and error-checked, so the models are now ready to be executed in GAMS by other team members.

Conclusions: There are still several stages of this project that need to be completed before conclusions can be drawn.  My contribution was to set up the interfaces, so that another member of the team could run the necessary code to produce a resulting TEP.  I was not given access to the necessary GAMS code to run the interface, and was also not given access to the IIT Comillas processing network, so all of my work had to be done on my personal laptop.  The completed and error-checked interfaces were emailed to another team member to run the code.  I am hopeful that I can be involved in reading and interpreting the resulting TEP whenever TEPES is executed.

Project output: I gave a brief presentation to Drs. Lumbreras and Galan at the end of my summer period, to discuss the next steps and show the progress that had been made.  I have attached this presentation in the email.

Personal Experiences: This program was my first experience traveling abroad, and I certainly enjoyed the time that I spent in Madrid and the surrounding area.  I was able to try many foods that I had not previously, which I will miss dearly.  Interacting with other members of the IIT Comillas team on a daily basis was also another perk of participating in this research program abroad.  The Madrid summer weather was impeccable for the duration of my time spent there, and I am very appreciative for having the opportunity to live in and experience a different climate for a short time.  I was able to travel for very cheap using the short-range and medium-range metro and commuter trains around Madrid, which allowed me to see much of the surrounding suburbs and country side.  Overall, the experience was very positive, and I am glad that I chose to participate this summer.

Project Title: Validation of a microscale model for flow over a wind farm with complex terrain

Participants: Daniel Tabas (Undergraduate Mechanical Engineering student, Johns Hopkins University); Jiannong Fang and Fernando Porté-Agel (EPFL)

Objectives: To improve the accuracy of local wind speed predictions over the Juvent wind farm in Switzerland by tuning a microscale model that takes into account the effects of complex terrain and surface roughness.

Activities: Design of simulation studies and analysis of results.

Conclusions: Domain size and meshing parameters have a significant influence on the performance of the model. Using the results from one simulation as boundary conditions for simulation over a second, smaller area (nesting) was not effective. Large errors that remained at three of the sixteen wind turbines were accounted for by including turbine wake effects in the simulation.

Project output (such as presentations made, posters, reports, papers): Poster presented at the Wind Energy Science Conference in Denmark at the start of the internship; final report in progress

Personal Experiences: I took advantage of Lausanne’s proximity to both Lake Geneva and the Swiss Alps, finding opportunities every weekend to go hiking, climbing, mountain biking, sailing, and more. I also enjoyed visiting other cities, experiencing traditional food and drink, and travelling to the Netherlands, Denmark, and Italy.

I worked to improve wind speed predictions at the hub of each turbine in the Juvent wind farm, located in the northwest of Switzerland. I tested several meshing parameters and developed Matlab codes to assess the accuracy of each simulation run. I explored the use of WindSim features that were not previously utilized such as nesting and wake modelling. Finally, I turned to data analysis and the testing of other simulation parameters in order to explain remaining errors.

Project Title: Comparison of Coastal Wind Speeds with Scanning Lidar and Satellite Measurements

Participants: Patrick Keyantuo (University of Puerto Rico, Mayagüez, Mechanical Engineering, USA) International Host: Ioanna Karagali, Technical University of Denmark – Risø (DTU)

Objectives: (1) Perform CNR range filtering applying a dynamic approach. (2) Reconstruct the wind field at each point using the wind direction information from a vertical profiling Lidar. (3) Compare the reconstructed wind field with instantaneous wind speed retrievals from a Synthetic Aperture Radar (SAR).

Activities: I was familiarized with the MySQL language to obtain the data with the measurements collected during the Reducing Uncertainty of Near-shore wind resource Estimates (RUNE) campaign on the west coast of Denmark. A traditional and a new dynamic filtering approach were employed to compare the variability of the measurements and increase data availability. The MATLAB software was used for the reconstruction of the wind fields for comparison with the SAR wind data.

Conclusions: Dynamic filtering has been employed and the reconstruction of the wind field and comparisons are in progress. Project output (such as presentations made, posters, reports, papers):A poster was presented at the Wind Energy Science Conference at the Technical University of Denmark, Lyngby Campus.

Personal Experiences: Staying in Denmark this summer has been a fascinating experience, and has inclined me to consider pursuing graduate studies in renewable energy. The summer school at the beginning of the stay provided a background for the research I would be performing. Participating in the WESC conference allowed me to interact and meet new people and also connect with the research I do at my institution. Overall, it was a great experience being immersed in a new culture with such a vibrant atmosphere, rich history and relaxed lifestyle. I was also able to visit other countries in Scandinavia, explore the city of Copenhagen and its surroundings.

Project Title: Using Lidar to measure rain droplet size distribution

Participants: Magdalene Koo,  Mikael Sjöholm, DTU

Objectives:  Determine the effects of precipitation on Lidar measurements Examine the differences in Lidar measurements due to different shot peening pellets

 Activities: To examine the effects of precipitation on Lidar measurements, I compared the data from two days with similar wind conditions but one day with precipitation and the other without.  These two days, January 31 and February 4, were chosen by obtaining and examining weather data from the area and making plots of the precipitation amount, wind speed, and wind condition from a nearby Met Mast and weather stations.  The Lidar data from these two dates were then taken and signal return from the blades was filtered out. From here, the effects of precipitation on Lidar signals can be determined. Shot peening pellets were examined as they are similar in size to rain drops, but more easy to control.  The consequences of different shot peening pellets on Lidar signals was examined by looking at the frequency and maximum distributions of the returned data.  Changes and differences in these parameters between experiments would reveal the size and speed the pellets were traveling at and thus show a possible effect rain droplets would have on Lidar signals.

 Project Outputs: Plots for spinner Lidar returns for different dates

Proposed conclusions: In the remaining week, I will be discovering different methods of looking at the data in attempt to uncover the underlying relationship between the Lidar signal and the weather.  This will entail examining the data each hour and comparing where the peak power return occurred and how it is distributed.  Differences in these two parameter between the two dates will provide a point of comparison, allowing conclusions to be drawn.  For the shot peening experiment, the analysis methods will be applied to experiments with different pellet material and size to examine the effects those factors have on Lidar signals. Understanding how these two correlate will then provide additional means to realize the relationship between rain drops and Lidar returns.

Personal Experience: I have thoroughly enjoyed my time here in Denmark, and have learned so much on wind energy.  Prior to this experience, I did not even know what a Lidar was and have come quite far to be able to analyze the data from them.  In addition, I was able to gain experience in the engineering field beyond the college classroom and have a taste of what comes after graduation, an opportunity I greatly appreciate.  I have also acquired an interest in the field of wind energy and further see the importance and practicality of green energy to make our world a bit cleaner. Beyond the academia, I have also loved living in Copenhagen and exploring Denmark.  From biking around the streets downtown to visiting giant wind turbines on the west coast, this experience was truly unique and I am very grateful to have had the opportunity to do this.

Project Title: Simulating Open Channel Flow over Textured Surfaces

Participants:  Khalil Castillo-Aponte, Dr. Murat Tutkun, Dr. Gerardo Carbajal, Dr. Luciano Castillo, in collaboration with Dr. Mikael Mortensen

Activities: The goal of investigation was to develop computational descriptions of experimental systems in an effort to increase supporting information related to experimentally based conclusions regarding observed flow patterns.

Project Outputs: The bulk of effort was applied to understanding the CFD software FEniCS and implementing FEniCS over computational meshes which correlated with key components of experimental surfaces. Computationally, the effects of cylindrically shaped, pillar arrays upon flow speed and pressure at the near wall and surrounding positions were tested and described for two dimensions. The results were considered in the context of the mathematical conclusions for flow derivative of experimentation on similarly arrayed and structured micro surfaces.

Proposed conclusions: Ultimately, the summer investigation provided a foundation for developing experimental surfaces computationally. Computational experiments tested pressure and velocity gradients surrounding two dimensional flow over variously arrayed cylinders. The results did not undeniably refute the mathematical explanations for experimentally observed flow behavior in regard to the two surface qualities tested computationally. Future work will add elements to the computational systems that increase correlation with experimental microsurfaces including divergent geometry across 3 dimensions. These expanded computational systems will provide more definitive conclusions on flow behavior and better support or refute the results of experimentally generated data from the same or at a minimum, more similar systems.

Personal Experience: the experience was a unique one both scientifically and socially. The opportunity to travel and work in different settings with different groups and organizations is always rewarding and promotes scientific as well as personal development.

Summer 2016

1: Project Title: Reserve Modeling in Dutch Day-Ahead and Balancing Electricity Markets

Participants: Robin Border Hytowitz, (Johns Hopkins University Graduate, Environmental Engineering),   Özge Özdemir, (international host)

Objectives: Model different types of reserve in the electricity market, including purchasing reserve in the spot market, contracting with generators, sharing across regions, allowing for different requirements and allowing participation from different types of generation

Conclusions: Contracted reserve using seasonal requirements for reach country alone produces the highest cost results while market-based, daily requirements for a region produce the best case scenario.

Project output: Paper in preparation, presentations upcoming

Personal Experiences: It was a wonderful return trip to Amsterdam, where I was able to reconnect with colleagues at ECN and experience the beautiful summer weather. I enjoying taking a bike trip to a nearby medieval castle and wandering around local outdoor markets.


2: Project Title: A Highly Resolved Large-Eddy Simulation of a Wind Turbine using an Actuator Line Model with Optimal Body Force Projection

Participants: Tony Martinez, (Johns Hopkins University Graduate, Mechanical Engineering), Richard Stevens (international host)

Objectives: To study the modeling of nacelle and tower in LES using Gaussian body forces.

Activities:  Visited University of Twente fo 3 weeks and went to the Torque 2016 conference in Germany to present a paper.

Conclusions:  The work form the conference has been presented and published.

Project output: Torque 2016

Personal Experiences:


3: Project Title: Reactive Power Extension to the DCOPF

Participants: Brent Eldridge,  (Johns Hopkins University Graduate, Environmental Engineering), International Host: Pablo Frias and Tomas Gomez

Objectives: Create a model to see if increasing amounts of intermittent renewable resources like wind will require better dispatch algorithms that solve voltage constraints optimally.

Activities: Extended the DCOPF model typically used in generator dispatch to include linearizations for reactive power and voltage.

Conclusions: The extended model works on smaller systems but more analysis needs to be done to scale it to larger networks.

Project output: A paper is in progress, and intermediate results will be shown at the 2016 Trans-Atlantic Infraday Conference at the Federal Energy Regulatory Commission in Washington, DC.

Personal Experiences: I traveled to Europe for the first time and had the opportunity to visit a handful of Spanish cities and neighboring countries while making significant progress towards my thesis proposal. I would definitely like to visit again if the opportunity presents itself.


4: Project Title: Multiscale analysis of turbulence with application to space-time correlation models

Participants: Perry Johnson (Johns Hopkins University Graduate, Mechanical Engineering), Michael Wilczek (lead international host). Other collaborators: Christian Lalescu, Theo Drivas

Objectives: Explore the underlying physics of turbulence relevance to the “random sweeping” hypothesis, which is used for constructing simplified models for space-time correlations in turbulent flow fields, particularly of interest to large-scale wind farm arrays.

Activities: Continue on-going collaboration by detailed analysis of numerical datasets from turbulence simulations. Resolve discrepancies in data analysis. Collaborate to interpret the results into publishable form.

Conclusions: Important discrepancies in the numerical data processing were resolved by working together closely on the details of the data analysis. Collaborative discussions led to consensus on the interpretation of the data.

Project output: Results will be published in archival journal.

Personal Experiences:  This was my first international collaborative trip to Europe. I was able to interact with other researchers at the Max Planck Institute interested in similar scientific topics. While there, I presented a seminar on some of my research results, and had many additional discussions for sharing research results and ideas.

5: Project Title: Extension of model-based receding horizon control of wind farms for secondary frequency regulation.

Participants: Carl Shapiro , (Johns Hopkins University Graduate, Mechanical Engineering), International Host: Johan Meyers (KU Leuven)

Objectives: Extend model-based receding horizon control of wind farms to include rotational inertia of turbine rotor, generator torque, and blade pitch angle

Activities: Considerable progress was made during the stay at KU Leveun. The wake model was extended to include power and thrust coefficient curves and nonlinear rotor dynamics. Adjoint equations, which are needed to implement optimal control techniques, for these extended equations were derived. Furthermore, modifications to the actuator disk implementation in the LES code were made to include thrust and power coefficient curves.

Conclusions: Progress demonstrated that the actuator disk approach can be used to model the rotor dynamics and include power and thrust coefficient curves. The ad joint equations can also be derived with the extensions to the model.

Project output: This project is still ongoing


6: Project Title: Offshore Wind Farm Layout Optimizer

Participants: Thomas Marge (Johns Hopkins University,  Applied Mathematics & Statistics) – International Host: Comillas University, Spain, Sara Lumbreras Sancho, Andrés Ramos Galán

Objectives: Integrate turbine placement into the offshore cable connection optimization model.

Activities: The original OWL model was adjusted so that the distance between rows and columns of turbines could be modified on each run. The cost of moving turbines closer together was approximated using Jensen’s Wake Model, which allowed for power production deficit calculations based on the reduced wind speeds.

Conclusions: More work has to be done to make the model robust.

Project output: (such as presentations made, posters, reports, papers): A publishable paper is in the works, a project report has been submitted to the project advisers

Personal Experiences: I traveled all across Spain, Switzerland, Italy, France, Germany, Portugal, Belgium and the Netherlands both before the project began, during weekend excursions, and the week following the project. I met and connected with students from all over the world, many of whom I hope to meet up with if I return to Europe.


7: Project Title: Performance of a coupled WRF-LES model of a wind farm over complex terrain 

Participants: Gabriel Mantegna (Johns Hopkins University, Environmental Engineering.), International Host: École Polytéchnique Federale de Lausanne (EPFL), Switzerland, Jiannong Fang, Farshid Kardan, Stefan Scholten, Fernando Porté-Agel (lead international host)

Objectives: Simulate the Juvent wind farm in Switzerland by coupling a large-scale meteorological model to a small-scale fluid dynamics model

Activities: Put together a system of code to create a coupled model

Conclusions: None yet (still working)

Project output: Created report and presentation for colleagues at EPFL, scientific publication in progress

Personal Experiences: Enjoyed the mountains and culture of Switzerland greatly.


8: Project Title: Buoy and Satellite Comparison

Participants: Daniel Campos (University of Texas/Dallas, Mechanical, Engineering), International Host: Technical University of Denmark – Riso (DTU), Ioanna Karagali, Ebba Dellwik, Alfredo Pena

Objectives: Study and compare significant wave height measured from satellite altimeters to significant wave height measured from buoy.

Activities: Obtain satellite data from ftp website and run it through a code to gather specific data points within a certain criteria in order to compare it to buoy data given to me.

Conclusions: The satellite altimeter is very accurate and can be used to gather measurements for future projects.

Personal Experiences: Explored Copenhagen, cities around it, and also visited Sweden during my stay in Denmark.


9: Project Title: Lidar Measurements and Comparisons

Participants:  Jessie Opella (Texas Tech University, Mechanical Engineering), International Host: Technical University of Denmark – Riso  (DTU), Alfredo Pena Diaz

Objectives: The goal of my project was to take the data from a lidar to calculate wind speed. The data was then compared in different comparisons to find the best way of analyzing the data. This information would be ultimately used to determine the validity of the lidar and the best way to reconstruct a wind field.

Activities: I wrote a Matlab code that did the entire process. First, I create a visual for all of the data in a time looped contour plot of the wind speed. Next, smaller sections of the entire scan were used to hopefully increase the accuracy of the calculations. These sections were then compared so the best size could be determined. Finally, the sections were also compared to another lidar at the same location to determine accuracy.

Conclusions: After I made all of my comparisons, I was able to calculate the correlation between the different sizes. However, due to time and other constraints I did not make any further analysis or definite conclusions on the data or calculation process.

Project output: At the end of the summer, I presented a short power point to the department.

Personal Experiences: Overall I had a wonderful time in Denmark. It was cool to live and adopt the lifestyle of another country. I enjoyed my time very much and would urge others to participate in this program.


10: Project Title: Modeling Euphemia with Minimum Income Condition

Participants: Christina Cho, (Johns Hopkins University, Environmental Engineering) – International Hosts: Carlos Batlle, Universidad Pontificia Comillas: Instituto de Investigación Tecnológica
Ignacio Herrero, Universidad Pontificia Comillas: Instituto de Investigación Tecnológica – Inés Rodes, Universidad Pontificia Comillas: Instituto de Investigación Tecnológica

Objectives: The goal of the project was to replicate the Spanish electricity market clearing algorithm to observe market behavior with a further investigation focused on the effects of wind power.

Activities: First, a simple auction-based market was modeled in GAMS. The minimum income condition (MIC) orders were then implemented into the social welfare optimizing GAMS model using a linearized form of the condition found through various literature review. Real market data was inputted into the model using Matlab functions and its results were compared with actual market results by looking at Market Clearing Price cleared demand quantity, and total withdrawn energy.

Conclusions: The model similarly replicates the real market when comparing market clearing price and cleared demand quantity of the modelled and actual results. Further studies can be done by analyzing the effects of wind on market behavior. Overall, the model proves to be a useful tool in exploring different aspects of the electricity market.

Project output: I wrote a final report detailing the process and findings of implementing MIC orders.

Personal Experiences: I am very grateful for the amazing opportunity to conduct research with other students and professors who are very passionate about their field. Everyone at IIT was very friendly and welcoming, and I enjoyed meeting new people every day. On top of that, staying in Madrid for two months was an amazing experience! I was able to explore many different cities


11: Project Title: Storage Allocation and Investment Optimization for Transmission Constrained Networks Considering Losses

Participants: Dean Yacar, Johns Hopkins University, Environmental Engineering, International Hosts: Diego A. Tejada-Arango, Sonja Wogrin

Objectives: Develop and expand on an existing model for storage allocation and investment optimization in order to test the effects of ohmic losses on the network.  Once completed this model was further expanded to a stochastic version used to model line outages within the network. This second model is still being developed in collaboration with the university.

Activities: In order to complete this project a literature review was done of several studies, participation in a GAMS workshop was completed, extensive work formulating and coding models into GAMS was done and now the results are being summarized in a working academic paper.

Conclusions:  Conclusions were reached regarding the dependence of the amount and type of storage was highly dependent on the location of demand centers throughout the network as well as the form of the generation at specific buses.   The effects of losses were also considered and compared and correlated to effects of congestion. The paper in preparation will provide quantitative results regarding these issues.

Project Output:  A paper is being prepared for possible submission for publication.

Personal Experience:  This was the best internship I have had my entire time at Hopkins.  I really enjoyed working on an interesting research project on a topic that I am passionate about.  Working in an international setting allowed me to gain a new understanding of how to deal with issues in the energy sector. All of the people I met and worked with at Comillas were very friendly and helpful which made the transition easy and comfortable. I was also able to take some Spanish classes part time at the university which helped me acclimate myself better to the new environment.  I highly recommend this program to other people and would gladly do it again myself if possible.

Summer 2015

1: Characterizing Wind Tunnel Surface Roughness

Participants: Max Marshall (Johns Hopkins University Civil Engineering Undergraduate, USA), Prof. Fernando Porté-Agel (PI, Wind Engineering and Renewable Energy Laboratory, EPFL, Switzerland) Vincent François-Charles Rolin (PhD Student, Wind Engineering and Renewable Energy Laboratory, EPFL, Switzerland)

Objectives:  The goal of this project was to further explore and implement a technique that Professor Porté-Agel had used previously, but not yet in the wind tunnel at the EPFL.  This is to place roughness elements along the main testing section of the wind tunnel, ideally to more accurately simulate a scaled atmospheric boundary layer.  The specific objective of this project was to observe how various surface element configurations influenced the mean velocity profile, and how these different profiles compare to real-world boundary layer conditions.  This is motivated by the general idea that the more accurately the boundary layer in the wind tunnel reflects the atmospheric boundary layer, the more accurate wind tunnel tests will be.

Activities:  The process initially consisted of literature review in order to become familiar with the necessary aspects of atmospheric boundary layer study.  Then, velocity profiles were taken at various locations in the wind tunnel, without extra surface roughness added.  These tests were performed by pitot tube mounted on a 3-axis traverse system.  Then, based on a similar technique used by Professor Porté-Agel previously, chains stretched spanwise across the tunnel were used to simulate roughness, with turbulence triggered by a picket fence near the beginning of the tunnel.  Different chain spacings were tested via pitot tube.  Then, the experimentally determined velocity profiles were fit to the standard atmospheric boundary layer logarithmic curve in order to obtain parameters to compare to real-world scenarios.

Conclusions:  According to the (somewhat limited) parameters offered by the mean velocity profile, various chain spacings could be used to generate scaled roughness lengths that correspond to useful real-world situations, like grasslands, plains, and crops, where wind farms are likely to exist.  However, before relying on this setup, it is advisable to examine the conditions with more sensitive equipment to characterize turbulent effects.

Project Output:  Preliminary results were presented via a poster at the WindFarms 2015 conference at KU Leuven in July.  Also, a MATLAB script was generated to enable efficient similar analysis in the future.

Personal Experiences:  I am enormously grateful for the opportunity to participate in this program.  Aside from learning about all sorts of interesting facets of engineering research, living in Switzerland for 2 months was an eye-opening experience. Lausanne is a beautiful city, with Alps looming in the distance over Lake Geneva.  I was able to travel all over Switzerland thanks to the always-reliable system of public transport, and visited France and Italy as well.  It was amazing to see the changes in culture, architecture and even climate that are packed into such a small area.  Also, I enjoyed the weeklong visit to Belgium for the WindFarms 2015 conference immensely, and was able to cap my trip with a backpacking adventure in the Swiss Alps.  The Matterhorn is incredible!  However, the best part of the trip was meeting so many interesting people all over the place.  Everyone at the EPFL was helpful, available, and friendly.  It was amazing to go out to lunch with a group of graduate students from all over the world, with a view of the Alps, and feel perfectly welcome.  The crowd of engineers and researchers at the conference were equally amicable.  I appreciate the efforts of everyone on both sides of the Atlantic that make this program possible.


2: Calibration of the EPFL Wind Tunnel Using Lift and Drag Generation on Known Airfoils

Participants: Martin Rich (Undergraduate Mechanical Engineering student, Univ. Texas Dallas,USA) Majid Bastankhah (graduate student EPFL, Switzerland), Fernando Porté-Agel (Lead International Host at EPFL, Switzerland)

Objectives:The objective of the experience was the re-evaluation of the data collection structure to ensure error generation was at a minimum. The collection of accurate data for lift and drag forces of NACA 0012, flat plate and a cambered airfoil provides information on the current status of the wind tunnel.

Conclusions: The data was collected successfully; a report on the findings was compiled and given to the EPFL WIRE group for understanding of the wind tunnel’s condition.

Project Output: We presented findings at the conference in Leuven Belgium as a poster.

Personal Experiences: I traveled and experienced the sights, culture and beauty of Switzerland and surrounding areas. I began teaching myself French, out of a desire to speak with more people. I made friends and was able to network with advantageous individuals.


3: Wind Production Scenario Generator 

Participants: Phoebe Yeo (undergraduate student at Johns Hopkins Universit, USA), Javier Reneses (professor at Universidad Pontificia Comillas ICAI in Madrid, Spain), Antonio Bello (PhD student at Universidad Pontificia Comillas ICAI in Madrid, Spain)

Objectives: Our goal was to expand upon a scenario generator of wind production to include data from Germany and also solar data.

Activities: I first learned the methodology developed by ICAI to predict wind energy production.  The methodology consisted mostly of time series analysis techniques.  I then adapted the methodology to past data given to me from Germany.  All experiments were done using MATLAB.  The scenario generator is meant to be one of many factors considered regarding electricity price forecasting.  I had to familiarize myself with the different methods of price forecasting, such as hybrid methods, through reading articles and research papers.

Conclusions: We were able to create a scenario generator for wind production in Germany.  The scenarios were per month over a span of a year.  We created different paths based on centiles and ran multiple trials in order to best imitate the volatility of the historical data.  We were then able to take averages and compare the results with the previous results obtained for Spain and Portugal.

Project Output: A poster was presented at the Windfarms 2015 Conference in Leuven, Belgium.  A preliminary draft for a research paper was composed.

Personal Experience: Before doing the summer internship, I had previously studied abroad in Madrid for 4 months.  It was great to be able to extend my stay and travel a bit more.  The graduate students at ICAI were extremely welcoming, and it was interesting to see the differences in the higher education system between Spain and the United States.  I enjoyed my time and already miss the food, the culture, and the people!


4: Optimizing Storage in a Two-Market Framework

Participants: Yong Kai Saw, (JHU undergraduate student, USA) Professor Sonja Wogrin (Comillas, Spain), Professor Dennice Gayme, Professor Ben Hobbs (JHU, USA).

Objectives: In this project, there is a storage operator and system operator both working within a fast and slow energy market. The purpose of this set up is to understand how the storage operator will allocate its power generation capabilities between each market, and how will the activities of the storage operator affect the system price. This project will look at scenarios of congestion demand fluctuation.

Activities: Based on operational and physical constraints, different formulations of a GAMS code were programed to obtain quantitative results. These GAMS formulations included: optimization, non-linear program (NLP), mixed complementarity problem (MCP), mixed integer program (MIP), and a bi-level programming problem.

Conclusions: The conclusions for both a perfect competition market as well as a market in which the system operator had market power are preliminary and still need to be reviewed and worked on.

Project output: Two virtual presentations were made to the WINDINSPIRE group, and one poster was made for a Wind Energy Conference in Leuven, Belgium.

Personal Experiences: Through the dedication and guidance from my mentors, I was able to learn different formulation methods to tackle an energy systems problem that is both relevant and important to societies around the world. By staying in Spain for the summer, I also gained a deeper appreciation for travel and exploration.


5: Optimization and Wake Modeling of Wind Turbines within Offshore Wind Farms

Participants: Christine Hamilton (Smith College undergraduate student, USA), Sara Lumbreras and Andrés Ramos (Comillas University, Spain), Charles Meneveau and Richard Stevens (Johns Hopkins University, USA)

Objectives: (1) To model the wakes of turbines in an offshore wind farm. (2) To linearize the power outputs projected by these models in order to put them into a linear programming optimization problem. (3) To modify a program that could optimize the electrical layout of a given offshore wind farm so that it could change and optimize the positions of the individual wind turbines, taking into account increased power outputs resulting from a larger spacing between individual turbines.

Activities: The wakes of wind turbines in a given offshore wind farm were modeled in Matlab using both the Jensen model, which assumes that wakes expand linearly, and a modified model that took into account interactions with the atmosphere as well. The power outputs generated from this model were used to develop a linearization estimate that could predict the power output for any reasonable distance between wind turbines. This linearization was then used to modify a GAMS model that optimized the electrical layout of a farm without changing the position of turbines. Optimization of the distance between turbines was added to this model.

Conclusions: Preliminary results from the GAMS model showed that, as expected, increased distances between turbines causes investment costs to increase and operating costs to decrease. Whether it is better to have turbines closer or further apart in a farm is dependent on the cost of land and materials, and the amount of power each turbine can put out. Some possible next steps for this project include testing the code to make sure it is accurate and adding more variables to the code so that the turbines can move in two directions and don’t have to stay in a grid shape if it is not the most optimal solution.

Project output: A poster on this project was presented at the WindInspire conference in Leuven, Belgium this summer. Additionally, a presentation was given during one of WindInspire’s monthly virtual meeting conference calls. A paper on this work may be forthcoming later this year or next summer.

Personal Experiences: Living in Madrid for two months this summer was an amazing experience! I was able to explore Spain and learn more about its language and culture. I met students and professors from around the world, and learned a lot about how research is done and what is involved in it. This was my first trip to Europe, and I hope to be back again soon!


6: Investigating the effects of wake meandering in wind turbine arrays.

Participants: Joel Bretheim, (Johns Hopkins graduate student, USA), Charles Meneveau, Dennice Gayme (Johns Hopkins University, USA), Fernando Porté-Agel (host Professor at EPFL, Switzerland).

Objectives: To use reduced-order computational models such as the Restricted Nonlinear model, in order to characterize the effects of wake meandering in extended wind turbine arrays.

Activities: Wake meandering is an important phenomenon to understand in the design of large wind farms. We have begun developing a new, low-cost computational tool to study this phenomenon. With this new approach, we hope to better understand the effects of meandering wakes on important turbine performance criteria.

Project output: This work will be presented at the APS DFD meeting in Boston this November (2015). Additionally, a presentation was given during one of WindInspire’s monthly virtual meeting conference calls.

Personal Experience: It was a pleasure to collaborate with the researchers at EPFL. They provided a great work environment and many stimulating conversations on various wind energy problems. On the weekends I hiked in the Alps and visited interesting Swiss locations such as the Large Hadron Collider at CERN and the cheese factory in Gruyères. All in all, I had a fantastic experience in Switzerland.

7: Methods for assessing power output in non-uniform onshore wind farms

Participants: Andrea Staid, (Johns Hopkins graduate student, USA), Pierre Pinson (Danish Technical University, Denmark).

Objectives: Maintain research connections, get feedback on current work and explore research directions.

Activities: Continued working on project analyzing wind farm power output.

Conclusions: Research still in progress, but gained some good insight from interactions with colleagues at DTU.

Project output: Presented at Leuven meeting. In process of writing up for journal submission.

Personal Experiences: I love Copenhagen and was so happy to spend some time there again. It was nice to see old friends from my previous visit and to meet new people and built connections.


8: Flexibility in the Dutch Electricity Market

Participants: Robin Broder Hytowitz, graduate student, JHU, USA, Dr. Ozge Ozdemir, Electricity Research Centre of the Netherlands, Prof. Ben Hobbs, JHU, USA

Objectives: Ro analyze a day-ahead and real-time model for the Dutch electric system to determine the quality and quantity of flexible resources needed 10 years in the future.

Activities: We ran additional simulations to include various degrees of collaboration between countries. The yearly data was analyzed across the course of 24 hours to determine which resources responded to changes in wind forecast error. Economic analysis of the day-ahead and real-time prices was also performed.

Conclusions: There will be need for additional flexible resources or better coordination of resources between day-ahead and real-time, and between neighboring countries.

Project output: Poster at the IEEE PES General Meeting and presentation at the INFORMS Annual Meeting

Personal Experiences: I was able to take several bike rides around the city, including a ride to Zaanse Schans on National Windmill Day. I enjoyed wandering around museums and eating local foods. Amsterdam is a wonderful city!


9: Large Eddy Simulations of Wind Turbine Wakes

Participants: Tony Martinez, (JHU, USA),  Mahdi Abkar (postdoc, EPFL Switzerland), Fernando Porte-Agel (host at EPFL, Switzerland), Ali EmreYilmaz (graduate student , KU Leuven, Belgium), Johan Meyers (Professor, KU Leuven, Belgium), Hamid Sarlak (postdoc DTU, Denmark), Robert Mikkelsen (DTU, Denmark), Jens Sorensen (DTU, Denmark).

Objectives: Performing a rigorous code comparison for simulating a wind turbine using an actuator line model and large-eddy simulations. Attending the wake 2015 conference in Sweden, and the wind farm meeting in Leuven, Belgium.

Activities: I visited several institutions to get started on the code comparison. This is now being worked on remotely with the collaborators, but the main technical issues were addressed during my stay. While visiting Switzerland I worked on an experimental validation of our computational code using experiments by the group of Prof. Fernando Porté-Agel.

Conclusions: We have set a deadline for the code comparison to submit a paper by the end of 2015. An initial paper was already was published for the Wake 2015 Conference which shows the first stage of the comparison with only 2 codes. We are currently working on a new paper for the experimental validation using the EPFL experiments.


10: Optimal control of wind farms for power tracking using simplified one-dimensional convection-diffusion equation

Participants: Carl Shapiro, (JHU graduat student, USA), Dennice Gayme, Charles Meneveau (JHU, USA), Johan Meyers (KU Leuven, lead international host, Belgium), and Pieter Bauweraerts (postdoc KU Leuven, Belgium).

Objectives: 1. Compare one-dimensional equation to LES simulations. 2. Learn about adjoint optimization for PDE constrained control problems. 3. Derive adjoint equations for one-dimensional equation. 4. Implement optimization program.

Activities: All of the objectives were pursued.

Conclusions: Given the simplicity of the one-dimensional model, the results compare very favorably to LES simulations, particularly the nonlinear model. Progress is being made on using the optimization methods for control, and comparisons with LES are expected shortly.

Project output: APS abstract submitted, and presentation made at the wind farm meeting in Leuven.

Personal experiences: Johan Meyers was a wonderful host during my stay. It was very nice to get to know the other students at KU Leuven as well.

Jump to top

Winter 2015

1 – Flexibility in the Dutch Electricity Market
Participants: Robin Broder Hytowitz, graduate student, JHU (@ ECN Jan. 15-31, 2015) Dr. Ozge Ozdemir, Electricity Research Centre of the Netherlands, Prof. Ben Hobbs, JHU

Objectives: The goal of this research stay was to continue and elaborate on analysis of the Dutch electricity market through the COMPETES model of the Netherlands and Europe

Activities: During my time at ECN, we were able to incorporate reserve requirements into the COMPETES model for day-ahead operation. We ran several test simulations in order to verify the use of reserves. In addition, we created wind generation scenarios using yearly historical data, and worked on scenario reduction techniques. Finally we analyzed several businesses cases from the COMPETES model to assess the amount of flexibility in future the Dutch market.

Conclusions: We observed that reserve requirements provided a great deal of operational flexibility compared to storage, demand side management, and a no-reserves case. Although costs were higher for the reserve simulations, there was significantly less load shedding. Further exploration of reserve requirements is likely needed, to ensure that there is enough reserve available but not too much to significantly increase costs or prices.

Project output: A paper is forthcoming from the project, and a conference paper from previous research has been submitted.

Personal Experiences: I very much enjoyed my time in Amsterdam and my colleagues at ECN. Simply walking through the city and enjoying small cafes was one of my favorite activities. The canals make the city quite majestic at sunrise and sunset, which is luckily late and early respectively during the winter.
2- A study of flexibility in customer response to load response signals
Participants: Li Zhao (PhD student), Mercedes Valles (PhD student at UP Comillas), Pedro Linares (professor at UP Comillas)

Objectives: Research customer response to load reduction signals in a pilot study conducted by a utility in Spain. Econometric methods were attached to make an empirical study.

Activities: Interpret variables available in the dataset provided by Spanish utility
Select variables needed Manage dataset to put information on the matrix format that we need Come up with possible models to use to test our hypothesis Possibly collaborate with others at Comillas who have worked on a similar task

Project output: We have come up with using logistic regression and autoressive models to estimate customer responsiveness and the degree of responsiveness. We also plan to estimate a supply of flexibility shown by customer as a function of price.

Personal Experiences: I was able to meet Nes PhD students at Comillas and other visiting scholars.
One visiting scholar will be coming to JHU, and I discuss her master’s thesis with her while I was in Spain since her thesis utilizes concepts from my two areas of focus/expertise. I learned Spanish jargon related to electricity demand as well as met with other people research a similar topic to demand response.

3- Transmission Expansion Planning Approximations and Strategic Storage Models
Participants: Roderick Go, Sonja Wogrin, Anya Castillo, Saamrat Kasina

Objectives: Expanding prior work on unit commitment approximations with generation investment to include transmissions investments on a network.   Code and test a strategic storage model in GAMS as an MPEC and MIP in preparation for testing.

Activities: Coded an transmission + generation expansion unit commitment model in GAMS. Developed appropriate code based on TRUC and p-TRUC approximations from Saamrat Kasina.
Debugged a strategic bilevel model in GAMS for investment and operation of storage facilities, formulating the original nonlinear problem as a mixed-integer using binary expansions and strong duality conditions. Began running preliminary case studies using nonlinear MPEC as compared to a simple, simultaneous investment + operations model.

Conclusions: For the transmission expansion planning model, we were able to successfully code the models in GAMS and run simple test cases; however, we are now working on developing appropriate data to use in case studies for publication.  After this visit, we had successfully debugged the nonlinear model to be able to run preliminary tests using stylized data. In these tests, we were able to see that the bilevel model did indeed produce different investment decisions than the more simple, simultaneous investment + operations model, in line with our expectations of strategic behavior.

Project output: Work is still in progress on both projects I worked on during this trip, but we are planning to prepare papers for publication by May.

Personal Experiences: During this visit, I was able to explore Madrid more with friends and colleagues I had met previously at Comillas during my visit in the summer. I was also more comfortable using my Spanish skills, communicating with new friends mostly in Spanish. Though the visit was short, I had very meaningful personal experiences exploring new parts of the city with new people, though I wasn’t able to explore places outside the city.
Jump to top

Summer 2014

1 – Comparison of Energy-Based and Power-Based Unit Commitment Using Simulated Marginal Costs
Participants: Alison Grady, undergraduate student, Smith College – (June-July 2014)
Germán Morales-España, graduate student, Universidad Pontificia Comillas
Prof. Andrés Ramos, Universidad Pontificia Comillas

Objectives: In my research this summer I focused on day-ahead unit commitment. My project was closely related to Germán Morales-España’s PhD thesis, which proposed a new power-based unit commitment model. The objective of my project was to compare the conventional, energy-based unit commitment model with this newly proposed power-based model, using marginal costs as the basis for comparison.

Activities: Using GAMS, I ran energy-based and power-based unit commitment models that met the same power demand curve. The power-based unit commitment met the instantaneous power demand at each moment, while the energy-based unit commitment met the averaged hourly energy demand. I used the solutions from these as a basis for a five-minute economic dispatch model that simulated real-time operation. I then compared the effectiveness of the two unit commitment models based on their marginal electricity costs. I also compared the marginal costs of the models with an added reserve margin equal to 2% of the demand. Additionally, I ran a set of experiments in which I removed the ramping constraints of the system. The ramping constraints limit how much the power output of a generating unit can change in one hour; they are inherently present in real generators, but not always considered in unit commitment models.

Conclusions: The power-based unit commitment performed much better in the marginal cost analysis, as the energy-based unit commitment was unable to meet demand in several instances and incurred high penalty costs for non-served energy. With an added 2% reserve margin, the results were very similar. Both the power-based and energy-based models had odd spikes in the marginal costs, and the energy-based unit commitment also had areas of very high marginal costs (equal to ± $1000/MWh) whenever there was non-served or surplus energy. When the ramping constraints were removed, all of the odd price spikes and almost all of the areas of non-served and surplus energy disappeared. This shows that the ramping constraints present in real generators are responsible for many of the energy shortages in real-time operation, so they should be considered in day-ahead unit commitment. Overall, these results confirm the conclusions in Germán’s thesis that conventional unit commitment is unable to meet demand, and a power-based model is needed for accurate power system operation.

Project Output: I have written one paper on this research that I plan to present at a conference sometime in the 2014-2015 academic year. I also plan to continue my research in the fall semester as an independent study, focusing on the non-served and surplus energy resulting from the two unit commitment models at varying levels of demand.

2 – The Evaluation of the EOP Solution as the Solution to the EP for Generation Expansion
Participants: Cecilia Lee, undergraduate student, Smith College at Comillas – (June-July 2014) – Adelaida N. Gomez, graduate student, Comillas Pontifical University
Objectives: “The equilibrium problem (EP) can be solved by several models, each with its own benefits and drawbacks. The two models that were studied are the EP solved by the diagonalization method (DM) and the equivalent optimization problem (EOP). The objectives included: 1) to compare the computation time of the EOP to that of the DM; 2) to evaluate the EOP solution as an approximation to the DM solution; and 3) to compare the EOP solutions of various number of system states representing the system to a given EOP solution that is deemed as a good approximation to the solution of the EP. ”

Activities: “During the preliminary stages of the research, I had to familiarize myself with the concepts of system states and generation expansion models. The research also required some background in game theory.
All computations were completed using GAMS and MatLab. A number of case studies were conducted to complete each objective mentioned above and to examine how the solution of the EOP changes as certain parameters of the model, such as different numbers of system states and gap value, are modified. ”

Conclusions: For a number of systems states, from 20 to 100, the computation time for the EOP and the EP were essentially equivalent as the difference was negligible. However, for a number of system states greater than 1000, the EOP computation time increased 15 folds, as the EP computation time was longer than two days for 1760 system states. Therefore, it was deemed that solving the EP for more than 1000 system states is not desirable. It is also unrealistic for a small system and computationally too heavy; it is more reasonable to run the EP with 760 system states or less.
Moreover, for the investment and production decisions, and the operation costs, the EOP solution is an adequate approximation to the EP solution. Comparing each of the solutions solved by the EOP with 24, 96, 168 and 264 system states representing the system to that of the EOP with 336 states yielded unexpected results. It was anticipated that the percent error of the solutions would decrease as the number of system states increased. However, the opposite occurred.

Project output: A research paper was written. Further research will be carried out during the 2014 fall semester.

Personal Experiences: “Madrid was a pretty awesome city. The friendly locals, tapas, calamares sandwiches, and jamon really enhanced my experience. The nightlife, although, for me, daunting, was quite a spectacle as well; the perfect time to people watch. From hipsters to businesspeople, locals and non-locals took to the streets, bars and night clubs. It was interesting to see that the nightlife was not reserved for only the young. What was even more interesting is that Spain seemed to run on its own time schedule, aptly named Spanish time – the locals eat lunch at 2pm and dinner at 10pm. I guess this is one reason it seemed to never get late in Madrid.
During the weekends, I got to travel around Spain. I visited Toledo, Segovia, Valencia, and Barcelona. One of my most memorable meals in Spain was in Segovia. Cochinillo is what the dish was called. All in all, I have acquired invaluable knowledge on my research topic as well as travel savvy. I am grateful that I was given this opportunity to research abroad.”

3 – Identifying the Ability of Lidar in determining Turbulence
Participants: Christopher Klement, undergraduate student, UTDallas,
Dr. Ameya Sathe, DTU, Risø – Roskilde, Denmark

Objectives: (1) Understand the similarities in Sonic Anemometers and RHI Lidar and compare the data from each site. (2) Determine the accuracy of Lidar simulations compared to real life Lidar

Activities: (1) Learned how to program in MATLAB in order to correctly extract, calculate, and compile wind speeds from the data provided. (2) Determine the rules around wind data and how to apply to the situation, involved; reading research papers, online help, and emailing collogues that originally took the data.

Conclusions: Greater understanding of wind energy methods and practices. Better understanding of research practices and ability to adapt to the task need of me. (I had previously had minimal MATLAB knowledge)

Project Output: I was able to provide Dr. Sathe with the program I wrote in order to extract and calculate both the Sonic and Lidar data.

Personal Experiences: I was able to completely gain and new appreciation for Wind Energy and the practices associated with it. During the summer I was able to travel a few times around Denmark and saw several interesting things. After my research I was able to travel around Europe and greatly enjoyed the sights.

4 – Generation Planning Modeling Trends
Participants: Cynthia Bothwell (JHU graduate student, at Comillas: January 2014)
Dr. Mariano Ventosa (ICAI,Universidad Pontificia Comillas)

Objectives: Our goal was to review current status of generation planning investment models and determine best practices and areas for improvement when incorporating higher penetrations of intermittent resources such as wind and solar. Also, determine future research priorities in generation investment modeling.

Activities: Met with various students and faculty at Comillas to discuss project and their perspectives on the problem. The scope of literature review was expanded per recommendations. The European perspective of planning and markets was discussed and compared to American processes. Also, a good foundation on similar activities in South America was received due to Comillas research efforts in South America.

Conclusions: The overview of current research determined impacts using different techniques than are currently in place in industry. The modified techniques would result in better companion resources for intermittent resources – lowering cost of operation. The main focus areas for improvements were determined and gaps in comprehensive portfolio analysis were discovered.

Project Output: A paper was written for and the work was presented at the International Association for Energy Economics International Conference in June 2014. The work was also presented at the July 2014 WINDINSPIRE meeting. A more extensive paper is in progress for submission to an industry journal.

Personal Experience: “Traveling to Madrid was very rewarding even for the short two week period. I was immersed in the local customs and traditions and gained respect for the history and culture. Each day my fundamental understanding of Spanish improved since few citizens in Madrid speak English. I enjoyed getting out and walking around the city, trying new foods and of course working hard with the Comillas IIT research group.”

5 – Development of a wall model for immersed boundary and line actuator methods
Participants: Kenneth Carrasquillo (UTD graduate student, at KU Leuven: June & July 2014) Prof. Johan Meyers (KU Leuven) Prof. Stefano Leonardi (UTD)

Objectives: (1) The focus of the visit to KU Leuven was to acquire knowledge on Large Eddy Simulation (LES) in order to implement the LES capability on the UT Dallas flow solver. (2) Implement the Smagorinsky model with Van Driest Damping. (3) Add a wall model based on the standard log similarity law along with a damping as proposed by Mason and Thompson.

Activities: Foremost, a thorough study of the basics of Large Eddy Simulation was performed, models such as the Smagorinsky model, Dynamic model and the modulated gradient model were considered. For a first step on including LES to the UT Dallas Flow solver, the Smagorinsky model was considered and added as a separate FORTRAN90 module. To simulate the atmospheric boundary layer a wall model was also added to the code.

Conclusion: The LES Smagorinsky model as validated using the widely known data obtained by Kim and Moin for a periodic channel flow with periodicity on the spanwise and streamwise direction, the test case was done using same parameters as Kim and Moin and all statistics were in perfect agreement. Later the model was added to a simulation with a wind turbine in a wind tunnel to compare with data from Krogstad. Oscillations that were observed previously on the flow field were automatically damped obtaining more accurate results in les computational time.

Project Output: The work done was presented on a KU Leuven group meeting organized by Prof. Johan Meyers. Results obtained from this code improvement will be used in two papers that are being written.

Personal Experience: “It was an exciting summer! The opportunity of being on the capital of Europe was amazing. I was exposed to and interacted with copious amounts of people coming from different cultures and this was, by far, the most rewarding experience of the summer. Being on such a convenient location in Europe enabled me to visit the historically rich city of Berlin and the beautiful and lovely city of Paris. Within Belgium I visited the breathtaking city of Bruges among other wonderful places such as Brussels and Namur. Watching the World Cup on Leuven along fans from every country that was in it was fascinating. At KU Leuven I had the pleasure of meeting, learning and exchanging ideas with great international students.“

6 – Open Source Energy and Momentum Tubes
Participants: Michael Howland (JHU undergraduate student, at Katholieke Universit, Leuven, Belgium June – July 2014)

Objectives: With the global effort to increase clean energy production capacity, wind energy, both onshore and offshore will experience increases in implementations, resulting in more wind farms with larger surface areas covered, as well as increased turbine density in a given surface area to conserve spacing. In an effort to optimize the transport of important properties such as mean flow kinetic energy and momentum in turbulent atmospheric boundary layer flow, an open source user defined MATLAB function is to be generated.

Activities: First, a general knowledge of CFD post processing was acquired through literature review. Professor Johan Meyers then provided a specific code which was used to generate the mean flow kinetic energy and momentum tubes for the publication: “Flow visualization using momentum and energy transport tubes and applications to turbulent flow in wind farms” Meyers and Meneveau 2013. This code was then studied to understand the general principles of generating the mean flow kinetic and momentum tubes. Production of the general code could then begin. In order to gain an idea for the different types of formats for the output of resolved data for post processing from CFD simulations, discussion with the advising professor and fellow students was vitally important. The umbrella function, called ‘generatetubes.m’ is run with inputs of velocity, Reynolds stresses, pressure, and mesh definition. The desired tube property is selected at this time. Other functions called generate the coordinate mesh grid of the tube according to streamlines extending upstream, as well as a function to generate more seed points upstream to account for a larger tube cross sectional area.

Conclusions: The user-defined function is able to take in any format of data for post processing. The mean flow kinetic energy tubes expressed the physical property of the vertical transport of Reynolds stress and therefore kinetic energy as the tube was traced upstream. This tool is an interesting visualization function, which can be used to help understand various flow properties through any wind farm CFD solution. This tool may be utilized for any size wind farm, or for a resolved fluid flow, which is not representative of a wind farm specifically. In the future, any CFD researcher may obtain this function and utilize it to generate the kinetic energy and momentum tubes for their experimentation in order to aid the literature.

Output: From this work, there was a short fifteen-minute presentation prepared upon departure from KU Leuven to inform the fellow students and faculty of the work. Additionally, the function, which generates kinetic energy, momentum, or mass tubes and cuts of cross section areas, will be placed online to be used on an open source basis.

Personal Experience: From a young age, cultural experience and travel have been an important part of my life. I was fortunate enough to be accepted into this wonderful WINDINSPIRE program which was fantastic academically, but also allowed me to meet, work with, and perhaps most importantly, befriend students and faculty from around the world. This internship was my first experience living alone, not to mention out of the country, but the friends I made at KU Leuven, and the nation of Belgium, made me feel right at home in the beautiful city of Leuven. While living in Europe, I had the good fortune to visit many cities in Belgium, as well as travel to Amsterdam, Paris, and London (all within two hours of Brussels). Between the new foods, new sights, new people, and new experiences, I had a wonderful time during this internship at KU Leuven.

7- LIDAR Data Analysis

Participants: Mark Yeatman (UTD undergraduate student, at DTU: June & July 2014)
Ebba Delwick (Senior Scientist at DTU)

Objectives: Develop software to help in the analysis of large data sets of LIDAR/Cloud Points. LIDAR data is used in all sorts of applications related to Wind Energy, most related to measuring wind speed. The data we worked with was terrain data in the form of LAS files, used in flow models and wind predictions.

Activities: While there is software available to handle LAS files, it is generally a small part of a larger geological software suite. There is open source software available as well, but it is difficult to use and difficult to add on more features for specific needs. In Matlab, we created classes capable of reading and writing LAS files, selecting subsets of data, and filtering the data. We also analyzed the quality of the data provided by a 3rd party vendor by comparing 3d renderings of the cloud points versus satellite images from Google Earth.

Conclusions: The vast majority results and conclusions of my time at Riso was the Matlab code I wrote. It was also concluded that the vendor data did not always correctly classify the data points, which results in the need to run the data through filtering algorithms to discover certain terrain features like buildings and rivers. We also found the MCC-Lidar algorithm is useful when attempting to identify buildings.

Personal Experience: “This trip was my first time leaving the United States for any significant period of time, and I quite enjoyed it. Without a doubt, it was a life changing experience. It was a particularly interesting trip for an undergraduate student to see sort of how the life of graduate student/researcher is. Life in Denmark is quite relaxed, and I found it quite pleasant. It was enlightening to see how life was different across the world.”

8 – Coordination and Planning of Electricity and Natural Gas Markets
Participants: Pearl Donohoo-Vallett, Post-Doctoral Fellow, JHU (@ ECN July 7-18, 2014) Dr. Ozge Ozdemir, Electricity Research Centre of the Netherlands, Prof. Ben Hobbs, JHU

Objectives: The broad goal of the project is to integrate an electricity market and natural gas market model to analyze the impacts of high wind penetrations. The specific objective of my trip was to integrate electric generation and electric transmission investment decisions into the model.

Activities: (1) Investment decision equations were formulated for integration into the model (2) Different power flow options for transmission investment were formulated for inclusion into the model (3) Irregular dispatch decisions were corrected

Conclusions: Preliminary runs indicated minimal investment in new natural gas infrastructure during high wind scenarios due to low coal prices and strong industrial/residential demand for natural gas.
Project output: Forthcoming!

Personal Experiences: Everyone at ECN in Amsterdam was exceptionally welcoming and friendly. Coming from a US view of cheap natural gas, the situation in Europe with expensive (and geopolitically unstable) natural gas supplies was an education.
I stayed in Haarlem, one stop away from Amsterdam on the Intercity trains. The pedestrian and cycling culture was a wonderful break from the traffic in and around Baltimore! I was converted to frites with mayonnaise, but I remain unconvinced that bread with butter and chocolate sprinkles is an acceptable lunch.

9 – Testing Unit Commitment Approximations on a Long Term Planning Model
Participants: Roderick Go (JHU DoGEE undergraduate student – June-July 2014
Dr. Sonja Wogrin (IIT, ICAI, Universidad Pontificia Comillas)
Saamrat Kasina (JHU DoGEE) and Dr. Ben Hobbs (JHU DoGEE)

Objectives: The primary objective of this project was to test how well four different approximation methods worked on a combined unit commitment and capacity expansion planning model. Due to the highly intermittent nature of wind energy, traditional approximations for the unit commitment problem do not work well in cases of high wind production and lead to poor long-term capacity expansion decisions. As such, developing and testing new approximations will aid planners in the industry in the future.

Activities: First, an integer-based, combined unit commitment and capacity expansion model was formulated and coded in GAMS based on previous work by Sonja and Saamrat. These models included the Tight Relaxed Unit Commitment (TRUC), the Tight Partially-Relaxed Unit Commitment (TPRUC), the traditional Load Block (LB), and the System States Representation (SSR). Using data from the Spanish network and electricity market operators, a one-week data set was created that roughly resembled the actual Spanish system. With this data set, the various approximation methods were tested against each other and the full hourly unit commitment model to determine how well they performed in terms of total system cost, electricity prices, investment into new capacity, and hourly energy production by generator type. After testing the models, we worked to improve the approximations to better capture the operational behavior we expected from the system based on the full hourly UC model.

Conclusions: Results show that the TRUC and TPRUC approximations perform very well; however, the TPRUC may suffer from long computation times for large problems. The LB model almost always does the worst due to the lack of chronology and startup and ramping constraints. Finally, the SSR model performed better than the traditional LB on which it is based; however, the SSR model suffered for problems with steep ramps. To address this issue, several tests cases were run, taking into account different system characteristics and by extending the model formulation itself to create stronger ramping constraints for key hours with steep ramps due to large changes in hourly wind production. These initial tests successfully improved the SSR model with respect to the results of the hourly UC model.

Project output: My work this summer was compiled into a report comparing the various approximation methods and the improvements made to the SSR model that we plan to develop into a conference paper in the near future after further testing.

Personal Experiences: Madrid is a great place to live, work, and relax. I enjoyed just spending time walking around the city, sitting in the cafes, and taking advantage of all the great cultural institutions. My colleagues at Comillas were wonderful, energetic, and more than helpful: it was great spending time after work getting to know them and the city better. I also really enjoyed the opportunity to travel during my stay: I was able to visit Toledo and Segovia for day trips, Barcelona and Valencia for weekend trips, and London for a few days at the end of the summer. I already miss the Spanish lifestyle and hope to go back soon.

10 – The role of pricing rules in generation investment with the presence of intermittent RES-E
Participants: Li Zhao, graduate student, Johns Hopkins University (June-July 2014)
Pablo Rodilla Rodriguez, Research Associate, Instituto de Investigación Tecnológica, Universidad Pontificia Comillas
Ignacio Herrero Gallego, graduate student, Instituto de Investigación Tecnológica, Universidad Pontificia Comillas

Objectives: Find the long-term capacity investment incentive differential caused by different types of day ahead market pricing rules when demand has a flexible component

Activities: 1) Build a central planner’s capacity expansion model to find an efficient mix as reference mix and generate a set of possible mixes. 2) Through mixes found in step 1, build unit commitment models to find hourly dispatch and market prices. 3) Calculate remuneration under different pricing rules. 4) Search for a break-even long-run competitive equilibrium to find the difference between the energy mix under the different pricing rules

Conclusions: In building our unit commitment model, capacity is altered to have flexible and inflexible components to estimate the effect of demand response on the influence of pricing rules. If capacity has a flexible component, unlike previously assumed, the long-run effects of different pricing rules is more narrow compared to when demand was assumed to be inflexible.

Project output: A working paper regarding the changes in results when allowing flexible capacity and demand response

11 – Measuring forces on model wind turbines in the EPFL wind tunnel
Participants: Michael Levidy (JHU undergraduate student – July-August 2014
Hosts: Prof. Fernando Porte-Agel (EPFL), Majid Bastankhah (EPFL)

Objectives: In order to extract energy from a fluid flow, it is necessary to understand the interactions between the flow itself and any collection device placed in its path. Given the increasing popularity of wind turbines in the alternative energy sector, a thorough investigation into the forces and torques acting on the tower of a turbine is extremely valuable. For this reason, an experimental study was planned to better understand the effects of various flow conditions on a wind turbine.
Jump to top

Summer 2013


1 – Maximum Value Wind Prediction

Participants:  Andrea Staid (JHU graduate student, at DTU: May & June 2013),  Prof. Pierre Pinson (DTU), Prof. Seth Guikema (JHU adviser)

Objectives: Our goal is to develop probabilistic forecast models to accurately predict the maximum value wind speeds during a given time period. High wind speeds can pose a large risk to offshore wind farms, affecting turbine reliability, safe maintenance operations, and operation planning.

Activities: Statistical models were developed and analyzed to assess maximum wind speed predictive accuracy. Different models and model training techniques were compared against each other and against traditional meteorological baseline forecasting methods to assess improvements in accuracy. Over three years of both forecasted and actual data were used for this analysis.

Conclusions: Our probabilistic predictions far outperformed simple, deterministic forecasts; having a distribution on the prediction gives a lot more information than a simple deterministic value. The best model chosen was a relatively simple linear model, which used a combination of relevant forecasted parameters as covariates. It was significantly more accurate than baseline methods.

Project Output: This work was presented at the WINDINSPIRE workshop on June 25, and it is anticipated that the work will also be submitted to a journal for publication. The paper is currently in progress.

Personal Exp: Denmark was a great place to live and work for two months. My host at DTU was fantastic and phenomenally helpful in an advisory role for this project. It was a great working environment, and it was refreshing to see how differently things are done in other institutions. Unfortunately, I have almost no Danish language proficiency to show from my visit, since nearly all of the other graduate students in my department were international and English was the common language used throughout the department.

2 – Using Blade Mounted Lidars to Measure Wind Velocities
Participants: Max Okinow, JHU undergraduate student, at Risoe DTU: June & July 2013, Dr. Jakob Mann, DTU Risoe Laboratory
Dr. Anders Teigtmeier Pedersen – DTU Risoe Laboratory

Objectives: Using two telescopes for a lidar mounted onto the blade of a wind turbine allowing for instantaneous measurements of angle of attack and wind speed. The goal of this research is to provide feed forward corrections to the turbine to increase the efficiency of the turbine by optimizing the pitch of the blades.

Activities: First, background on lidars was acquired. A lidar is a remote sensing system that measures the distance to a particle. The theory of using them to measure wind speed is that small particles are carried with the wind. These particles are light enough that they can be considered to be moving at the same speed as the wind. The light is then reflected off of these particles and the lidar measures the Doppler shift of the backscatter. There are two types of lidars, a continuous lidar as well as a pulse lidar system. This project deals with continuous lidars. The continuous lidar system emits laser light that is focused a specific distance from the emitter, which affects the sensitivity of the measurement. However a pulsed system uses the time of flight of the laser path to determine the range at which the signal is from. The following diagram demonstrates a basic continuous lidar system.
The dual-telescope wind lidar consists of two small telescope units with 1” optics connected to a modified ZephIR 300 continuous-wave Doppler lidar through fiber optic cables. The ZephIR lidar is installed in the spinner of a Vestas NM80 turbine and the telescopes are mounted on either side of one of the blades 16.2 m from the center of rotation. The two telescopes’ beam axes converge and intersect 5m in front of the blade along the chord extension line, where the focus beam lies. An optical switch operating at 10Hz is used to alternate readings between the telescopes.

Conclusions: The double mounted blade lidar is a functional remote sensing system in regards to the expected wind speed from the scada compared to the lidar readings. On average the lidar readings are larger than the scada readings. This can be explained by the scada being behind the blades and reading the wind speed after energy has been extracted from the wind. When resolving the wind speed taking yaw into account, this becomes more evident. One caveat is that due to the nature of the analysis removing ground spikes can create a situation where much of the data is unusable from one or both of the telescopes. This means that the system itself seems to be unreliable as a mode of measuring the wind speed. The angle of attack measurements fall in the expected range of values based on the literature. However further comparison testing should be done to figure out which method of calculating the angle of attack is more accurate. This should involve pitostatic tubes in concert with the blade lidars to have some sort of control and comparison testing.

Personal Experience: Traveling is a passion of mine. I love seeing other cultures and enjoying the food of other people. This trip gave me the opportunity to visit several other countries in Europe including the Netherlands, Belgium, Norway, Sweden, and of course Denmark. I have thoroughly enjoyed exploring the small portions of each country that I visited in my time here. Copenhagen itself is a city with filled with activities for people who are looking for it. I was able to run everyday around the lakes in the middle of the city. Also there are plenty of open air concerts to visit. One of my favorite activities was wakeboarding in the park near Christianshavn. Sometimes it was fun to sit in the parks and drink with some friends (a typical Danish activity). It was so nice to be able to go to a park and play soccer against the locals. It helped me to fill my weekends as well as allowed me to meet some residents of the city. We were took advantage of a couple of sunny days, which can be rare in Copenhagen over the summer, to visit the nearby beaches and even went swimming a little. I have truly enjoyed my time living in Copenhagen. Life is much slower and relaxed than what I am accustomed to. Working hours are somewhat flexible and I was astounded by the sudden disappearance of most of the staff taking their mandatory summer vacations. It was very weird to work in an almost empty office compared to the full offices of my previous summer internships. Another interesting aspect of Copenhagen is the plethora of people biking and public transportation options. I really liked how easy it is to get around both Copenhagen as well as Europe in general. Professionally I have learned a significant amount about wind power, wind turbines, and lidars. Additionally I have become more proficient in matlab during my time here, which will definitely be helpful next year. It was enlightening to attend both the summer conference. It allowed me to view my work in a larger context and introduced me to other topics being researched in the field. The summer school at Risø gave me the opportunity to understand lidars better, which was really helpful to figure out where my project fit in the scope of lidars and remote sensing measurements. I really appreciated these experiences, which helped me to understand the problems facing both my project and the field of wind energy in general. Hopefully my work will help the team with analyzing data better for future projects.

3 – Dispatch of Spanish generation plants with respect to the optimal spillage of renewables
Participants:  Michael Peven, undergraduate student, JHU, at Comillas June & July 2013,  Dr. Sonja Wogrin, Universidad Pontificia Comillas, Prof. Andres Ramos, Universidad Pontificia Comillas,

Objectives: (1) Learn how to implement optimization problems in GAMS, and then how to formulate a unit commitment model to establish the optimal dispatch of generation for Spain’s electricity market in a few future high-renewable penetration scenarios.

Activities: (1) Learn how to use the GAMS software. (2) Research into unit commitment models and the Spanish electricity system. (3) Create hourly profiles for the daily fluctuations that occur with wind and solar energy units along with an hourly demand profile. (4) Formulate a model and equations in order to optimize the dispatch of generation units with respect to a cost function

Conclusions: (1) Determined commitment schedule for the Spanish electricity system. (2) Realized importance of wind power in high-penetration future scenarios and how much operating reserve would be needed because wind output is difficult to predict. This shows a need for either better forecasting or for load balancing reserves like pumped-storage hydro systems.

Personal Experience: I had an amazing time, I was able to learn some spanish during my stay, and I especially enjoyed trying new types of cuisine. It was fun playing pickup soccer games at the university, but the most memorable part was at the end of the stay, when I was able to go to Pamplona for the San Fermin festival and run with the bulls.

4 – Wind Farm Optimization: Micrositing and Electrical Layout
Alex Caffee, JHU undergraduate student, at Comillas June & July 2013.  Prof. Andres Ramos, Comillas Pontifical University
Dr. Sonja Wogrin, Comillas Pontifical University Dr. Sara Lumbreras, Comillas Pontifical University

Objectives: (1) Optimize OWF layout by considering micrositing and electrical layout effects 16
simultaneously. (2) Maximize wind power output using a linear model dependent on turbine spacing. (3) Minimize electrical layout costs using OWL. (4) Implement new combined model in GAMS (General Algebraic Modeling Systems). (5) Test model on Barrows Wind Farm to check validity.

Activities: (1) Learned programming in GAMS and worked on a number of real life examples. (2) Investigated the wind and electricity industry to understand objectives and constraints. (3) Created Excel file to calculate power output based upon Katic model for a wind farm with different spacing between turbines. (4) Modeled total power output of a wind farm based upon the spacing between turbines and verified its results with the Excel file

Conclusions: (1) Understood electricity industry and the problems facing it today, especially in regards to renewables. (2) Successfully learned programming in GAMS and applied it to a real problem. (3) Applied the Katic model to understand the tradeoff between wind spacing and total power output. (4) Developed a working model for total power output using the Katic model using Excel and GAMS. (5) Developed an interest in the wind energy industry and the desire to continue my research into the upcoming year.

Project Output: Poster presentation for the WINDINSPIRE Conference in Copenhagen, Denmark

Personal Experiences: (1) Increased my Spanish language level to Intermediate through speaking to locals, colleagues, etc. (2) Explored Spain and learned about the culture. (3) Developed a passion for Europe and the lifestyle. (4) Traveled to Morocco due to its close proximity to Spain.

5 – Proper Orthogonal Decomposition analysis of LES of the NREL 5MW rotor
Participants:  Claire VerHulst, graduate student, JHU at DTU June 2013.  Prof. Robert Mikkelsen, DTU,  Prof. Jens Sørensen, DTU Prof. Charles Meneveau, JHU

Objectives: The primary objective of this visit was to build on relationships with several researchers in the field of wind energy at the Technical University of Denmark (DTU). To best match our complementary skill sets, we wanted to perform a POD analysis on high-resolution data of a single turbine wake computed by the researchers DTU. With insight from the wind turbine experts at DTU, this analysis will hopefully lead to improved understanding of the dynamics of vortex breakdown and wake recovery which could facilitate better placement of wind turbines and improved prediction of their future performance.

Activities: During my time in Denmark, I presented my work at the three-day International Conference on Aerodynamics of Offshore Wind Energy Systems and Wakes (ICOWES) where I met a number of researchers from Denmark, the Netherlands, Norway, and England. I learned about a wide range of topics related to wind energy including analytical modelling of wake losses, site observations using radar or lidar, and optimal placement of wind turbines in a large array. I also presented at an all-hands meeting of the PIRE grant collaborators where I met several researchers who study the impact of wind energy on the electrical grid and on energy markets. As part of this meeting I also had the opportunity to attend my first grant-writing brainstorming session which was enlightening and engaging. After these couple weeks of conferences and meetings, I began work with Robert Mikkelsen and Jens Sorensen from DTU. We settled on a simulation case, and they generated a small dataset: a time-series of velocity fields surrounding a single wind turbine. The remainder of my time in Denmark was spent working with this data and performing the POD analysis.

Outcomes: The collaboration with Robert Mikkelsen and Jens Sorensen of DTU is only just beginning. I have completed a preliminary POD analysis on the small dataset provided during my time in Denmark, and have identified the most energetic flow patterns in the wake of the turbine. We are now preparing to supplement this small dataset with more realizations of the flow to ensure analysis convergence. Once we are confident that we have identified the energetic flow structures correctly, we can study how they relate to vortex breakdown and wake recovery. We expect to present our results at the 66th annual APS Division of Fluid Dynamics conference in Pittsburgh in November 2013. We also look forward to publishing these results in journal article at some point in the future.

Personal Experiences: My trip to Denmark was a pleasant experience. During my time there I rented a room from a Danish woman and her daughter, so I had the opportunity to hear the language every day and to learn some of the culture. The city of Copenhagen was beautiful, and I enjoyed many long runs through the parks and downtown streets in the early mornings. I traveled to Roskilde and Helsingor and visited many of the museums downtown on the weekends. Overall, I felt a sense of community that I expect is largely a result of the fact that Denmark is a small and relatively homogeneous country compared to the US. I was also very impressed with the cycling infrastructure – that’s something we could certainly use here in the States.

6 – Sensitivity Analysis of the Approximation of Bilevel Generation Expansion Equilibria
Participants:  Adrienne Horne, undergrad student, Smith College, at Comillas June & July 2013
Dr. Sonja Wogrin, Comillas Pontifical University, Prof. Andrés Ramos, Comillas Pontifical University

Objectives: The primary objective of this visit was to build on relationships with several researchers in the field of wind energy at the Technical University of Denmark (DTU). To The objective of this project was to introduce myself to optimization and equilibrium problems used in power generation planning and become comfortable with modeling in GAMS in order to successfully analyze the results. These new skills would enable me to analyze the effectiveness of the single-level equilibrium generation expansion model developed by Sonja Wogrin when taking into account a larger number of generation companies and asymmetric parameters.

Activities: By first familiarizing myself with convex optimization and the GAMS programming language, I was then able to develop my own code for a market equilibrium problem, optimization problem, and complementarity problem on paper and then in GAMS. Able to understand the major concepts of bilevel equilibrium modeling, I coded the market equilibrium bilevel approximation problem in GAMS. I then streamlined the program and extended the existing approximation to take into account multiple players and asymmetric parameters. At this point, I have a functioning GAMS program that allows the modeling of multiple players and asymmetric parameters. I hope to continue my work with this program into the Fall so that I may perform a complete sensitivity analysis with parameters of the number of generation companies, demand and wind data, cost data, and strategic spot market behavior.

Outcomes: I have already presented a poster of these aforementioned objectives and future plans at the WINDINSPIRE workshop at DTU. But, after completing this analysis, it will be possible to assess the effectiveness of the approximation and complete a paper on these findings.

Personal Experiences: I really appreciated that this research opportunity allowed me to learn more about tools for power generation planning and modeling and to experience a new culture at the same time. Working at Comillas University allowed me to meet PhD students from all over the world. At lunch I was able to talk to friends from Serbia, Iran, Brazil about politics, choosing a career path in engineering, and cultural differences they’ve experienced travelling and living in Spain. It helped me think more about what I want for my future studies and put my views of the United States and relations with other countries in perspective. I also appreciated the chance to travel to Copenhagen for the WINDINSPIRE workshop at Danish Technical University. It was very inspiring for me to listen to lectures from so many experts in different fields pertaining to wind energy and power generation planning. I have never had such an opportunity to learn about so many different research paths and careers. It was very exciting for me to speak with these professionals and experts to hear about their experiences and think about possible future collaborations. For these reasons, I feel this internship opportunity was a really valuable experience.
Jump to top

Comments are closed.