Electric Vehicle Charging Station Location using Fuzzy Optimization

Electric vehicle charging station location is a basic problem in integrating electric vehicles in electric power systems. An electric vehicle plugged into the electric distribution system may absorb or produce active and/or reactive power [1-4] depending on the need of the electric power system, Table 1 below.

Electric vehicle charger operating modes [4].

When finding the location of EV charging stations in order to support the electric distribution systems, the cost of EV charging, the impact on distribution system losses and voltage profile of the system are parameters needed to be considered. These variables are to be looked into when plugged in EV is either acting as a generator or a load given a system demand level.

Recent studies have solved this EV charging station location problem. In [5], a mixed integer programming solution was developed with site accessibility, local jobs and population densities and trip attributes as main constraints. A genetic programming approach is utilized in [6] for simulation of electric vehicles on a real map of a European city where the optimal solution of the charging infrastructure is derived based on mean trip times of electric vehicles. A two step procedure is proposed in [7] where the authors included environmental factors and service radius of EV charging stations in the screening first step and built a modified primal-dual interior point algorithm (MPDIPA) for optimal sizing of EV charging stations with the minimization of total cost associated with EV charging stations to be planned as the objective function with losses and voltage profile included in the problem. Reference [8] introduces an optimization process for sizing and siting of EV charging stations, modeling the charging demand and the structure of road network to where the solution approach was graph theory. Level 1 and level 2 charging stations are discussed in [9] and how to allocate them for residential EV users using simulation-optimization strategy.

Recent studies do not consider uncertainties and imprecision which can be captured using fuzzy optimization. Fuzzy set theory can provide a simpler yet powerful solution for allocating EV charging stations in electric distribution systems. The Civanlar test system [11] will be utilized for the study and assuming that capital investment of the EV charging station is the same for all distribution system candidate nodes while considering time of use (TOU) electricity tariff, distribution system losses and voltage profiles.

References

[1] Chenye Wu, Hamed Mohsenian-Rad, and Jianwei Huang, “PEV-based Reactive Power Compensation for Wind DG Units: A Stackelberg Game Approach”, in Proc. of the IEEE Conference on Smart Grid Communications (SmartGridComm’12), Tainan City, Taiwan, October 2012.

[2]   Chenye Wu, Hamed Mohsenian-Rad, Jianwei Huang, Juri Jatskevich, “PEV-Based Combined Frequency and Voltage Regulation for Smart Grid”, the 3rd IEEE Innovative Smart Grid Technologies Conference, Washington DC, Jan 2012.

[3]    M. Kisacikoglu, B. Ozpineci, L. M. Tolbert, "V2G Reactive Power Compensation Using a PHEV Bidirectional Charger Interface Rated at Level 1, 2, and 3 Charging Standards," IEEE Energy Conversion Congress and Exposition, Atlanta, Georgia, Sept. 12-16, 2010.

[4]     M. Kisacikoglu, B. Ozpineci, L. M. Tolbert, "Examination of a PHEV Bidirectional Charger System for V2G Reactive Power Compensation," IEEE Applied Power Electronics Conference, Palm Springs, California, Feb. 21-25, 2010, pp. 458-465.

[5]    Chen, T. D., et al, “The Electric Vehicle Charging Station Location Problem: A Parking-Based Assignment Method for Seattle”, on-line: http://www.caee.utexas.edu/prof/kockelman/public_html/TRB13EVparking.pdf

[6]    Hess, A. Et al, “Optimal Deployment of Charging Stations for Electric Vehicular Networks”, on-line:http://conferences.sigcomm.org/co-next/2012/eproceedings/urbane/p1.pdf

[7]    Liu, Zhipeng, Wen, F. and  Ledwich, G. F. , “Optimal Planning of Electric-Vehicle Charging Stations in Distribution Systems”, IEEE Transactions on Power Delivery, Jan. 2013, Vol. 28 , Issue 1.

[8]    Jia, L., Hu, Z., Song, Y., Luo, Z., “Optimal siting and sizing of electric vehicle charging stations”, 2012 IEEE International Electric Vehicle Conference (IEVC), 4-8 March 2012

[9]    Xi, X., et al, “Simulation-Optimization Model for Location of a Public Electric Vehicle Charging Infrastructure”, on-line:http://www.ise.osu.edu/ISEFaculty/sioshansi/papers/charge_infra.pdf

[10]“Siting Electric Vehicle Charging Stations”, on-line: http://www.sustainabletransportationstrategies.com/wp-content/uploads/2012/05/Siting-EV-Charging-Stations-Version-1.0.pdf

[11]On-line: http://ee.sharif.ir/~e_hajipour/index_files/Test_Cases_Data.pdf

Voltage Stability Impact of Electric Vehicles

Reactive power consumption or injection is a promising feature out of electric vehicle charging. In [1] and [2], the authors present the analysis of the  reactive power control capability of an electric vehicle charging system to support the power system. Reactive power support and economics from the electric vehicles were discussed in references [3] and [4].

The electric vehicle charger follows the modes of operation [1]:

Electric vehicle charger operating modes [1].

Given these modes, analyzing power system voltage stability with electric vehicle charging would be needed. PV and QV curves will be helpful to assess the impact of the different charger operating modes as well as transient voltage stability simulations. This would be a welcome addition to the increasing literature of vehicle to grid (V2G) especially when the grid operator coordinates a large fleet of electric vehicle with the power network.

Does the electric vehicle charging mode provide increased power transfer in terms of static voltage stability? Does the operating mode of an electric vehicle charger gives a better voltage recovery during transient periods?

These research questions can be analyzed by modeling electric vehicle charging operating modes integrated in a power system test case.

References:

1. M. Kisacikoglu, B. Ozpineci, L. M. Tolbert, "V2G Reactive Power Compensation Using a PHEV Bidirectional Charger Interface Rated at Level 1, 2, and 3 Charging Standards," IEEE Energy Conversion Congress and Exposition, Atlanta, Georgia, Sept. 12-16, 2010.
2. M. Kisacikoglu, B. Ozpineci, L. M. Tolbert, "Examination of a PHEV Bidirectional Charger System for V2G Reactive Power Compensation," IEEE Applied Power Electronics Conference, Palm Springs, California, Feb. 21-25, 2010, pp. 458-465.
3. Chenye Wu, Hamed Mohsenian-Rad, and Jianwei Huang, “PEV-based Reactive Power Compensation for Wind DG Units: A Stackelberg Game Approach”, in Proc. of the IEEE Conference on Smart Grid Communications (SmartGridComm’12), Tainan City, Taiwan, October 2012.
4. Chenye Wu, Hamed Mohsenian-Rad, Jianwei Huang, Juri Jatskevich, PEV-Based Combined Frequency and Voltage Regulation for Smart Grid, the 3rd IEEE Innovative Smart Grid Technologies Conference, Washington DC, Jan 2012.

Dynamic Models and Simulations for Reduced and Approximate Philippine Major Island Power Grids

Since developing the power flow models of Luzon, Visayas and Mindanao, one step forward in these projects is to provide dynamic modeling of the generators, exciters, governors, etc.

I followed the references [1-3] for assuming models for each generation considering fuel types. Also, combined with these good sources, PowerWorld provides default data for the dynamic models including the generic wind generation dynamic models (for NorthWind generation, north of Luzon) and loads (motors and discharge lighting, etc).

For generation using diesel as fuel, I initially modeled the machine as GENSAL but WECC has indicated to use GENTPJ instead for reasons cited in reference [4].

As I'm using PowerWorld, I made advantage of the auto correction of dynamic data and proceeded with the validation of models.

As mentioned in other posts, I simulated flat runs and had the models respond as expected. The following plots are simulated three-phase faults where fault clearing time is in accordance with the Philippine Grid Code and assuming a single-line contingency.

Bus Voltages Plots for Fault on Balintawak 230 kV, tripping Balintawak-Araneta 230 kV Line (Luzon)

Rotor Angles Plots for Fault on Balintawak 230 kV bus, tripping Balintawak-Araneta 230 kV Line (Luzon)

Bus Frequency Plots for Fault on Lugait 138 kV bus, tripping Lugait - Tagaloan 138 kV Line (Mindanao)

Generator Speed Plots for Fault on Lugait 138 kV bus, tripping Lugait - Tagaloan 138 kV Line (Mindanao)

Various Plots for Fault at Quiot 138 kV bus, tripping Quiot-Banilad 138 kV Line (Visayas)

I'm planning to write a full paper on this work and if you are interested in the models or collaborate with me, drop me a message at ebcano@gmail.com.

References:
[1]    IEEE Recommended Practice for Excitation System Models for Power System Stability Studies, IEEE Std 421.5-1992
[2]    IEEE PES Working Group, Hydraulic Turbine and Turbine Control Models for System Dynamic, IEEE Transaction on Power System 7 (1992) 167-174.
[3]    Dynamic Models Package Standard 1. Available: http://www.energy.siemens.com/hq/pool/hq/services/power-transmission-distribution/power-technologies-international/software-solutions/Dynamic_Models_Package_Standard-1.pdf
[4]    Additional Information on GENTPJ Model. Available: http://www.wecc.biz/library/WECC%20Documents/Documents%20for%20Generators/Generator%20Testing%20Program/gentpj%20and%20gensal%20morel.pdf

New England 39 Bus Test System

The New England 39 bus test system is a power system test system usually utilized for dynamic simulations test and research. It is believed that this is an actual equivalent system of the New England grid in 1960s [1].

The data I utilized here comes from [2], where the power flow and dynamics data were given in PSS/E v29 format which were loaded into PowerWorld. The data has 1 kV base voltage in all buses which I changed to 345 kV to reflect the New England system voltages. In the PowerWorld's transient stability data validation (this is very cool!), I accepted the corrections identified in the dynamic data, mostly are time constant depending on the time step being studied.

Normally in stability simulations, it is imperative to run a no fault simulation or what they call flat run to verify that dynamic models are behaving in a manner without disturbance thus expecting flat plots of parameters.

Generator angles for no fault simulation.

A stub fault is another practical test if the response of the dynamic models is correct for a simple and fast fault disturbance. Here are example plots from a stub fault at bus 1 at 1.0 seconds and cleared after 0.1 seconds without tripping any line.

Generator angles for stub fault simulation.

Generator speed for stub fault simulation.

Bus frequency for stub fault simulation.

Bus voltages for stub fault simulation.

If you want the New England 39 bus test system, email me at ebcano@gmail.com.

References:
[1] Power Systems Test Case Archive. Available on-line: http://www.ee.washington.edu/research/pstca/dyn30/pg_tcadyn30.htm
[2] Pablo Ledesma, New England Test System, IEEE 39 Bus System, 10 generators, in PSS/E format (version 29). Departamento de Ingeniería Eléctrica Universidad Carlos III de Madrid. Available on-line: http://electrica.uc3m.es/pablole/new_england.html

Cost Allocation of SPS Service Using Cooperative Game Theory

Power systems planning and operations are usually defined by N-1 criterion. This means that in an event of a single contingency, no remaining connected transmission elements will be thermally overloaded, no bus voltage will be outside of acceptable limits, no system interface limit is violated, and system stability is maintained.

Special Protection Systems or SPS are widely utilized for increasing power transfer in transmission systems at the same time respecting security constraints [1]. SPS applications usually are generation rejection schemes (GRS), line/transformer transfer tripping and load shedding. GRS are designed to mitigate overloading of a transmission line or lines after an N-1 contingency near the vicinity of a generating plant or are employed to arrest increasing dynamic oscillation which may lead to unstable system conditions. Without the GRS, generation output is curtailed to satisfy the N-1 security criterion. With the GRS, the output of the generation is increased thereby increasing power transfer. Further, GRS also mitigates or delays the possibility of transmission expansion or investment due to transmission capacity constraints.

In a locational marginal pricing based electricity market, curtailment of generation (without GRS), specifically of a cheap generation due to the security N-1 criterion can be considered as transmission congestion. Transmission capacity limitations impede the generation output thereby decreasing the profit opportunity of the generation company (GenCo).  If a GRS is installed for this GenCo, the output of the GenCo is increased and thus there is a clear benefit for the GenCo in terms of profit. When transfer capability is limited, without GRS, the profit of a transmission owner (TO) is decreased due to less power wheeling charges. With the GRS, wheeling charges increase as a consequence of the added power transfer. This premise is the same with the electricity system and market administrator, called independent system operator (ISO), since the ISO charges for cost-based services including scheduling, system control and dispatch.  For the demand side, when generation output is curtailed due to congestion, without GRS, the resulting nodal prices at the demand’s location maybe higher than when a GRS is in place to increase generation output from a cheap generation.

GRS installations have embedded cost and actual service cost [3]. Since electricity market participants have various benefits in having a GRS installation, the cost of the SPS/GRS service must be allocated among the participants. Cooperative game theory [4-5] can be utilized in allocating fair cost on the beneficiaries of the SPS service.

The PJM 5 bus test system [6], shown Figure 1, is to be utilized as an example for the application of cooperative game theory in sharing the SPS service cost among power system organizations.

Figure 1. PJM 5 bus test system.

References:

[1]     W. Fu, S. Zhao, J. D. McCalley, V. Vittal, N. Abi-Samra, “Risk Assessment for Special Protection Systems,” IEEE Transactions on Power Systems, vol. 17, no. 1, pp. 63-72. February 2002. Available: home.eng.iastate.edu/~JDM/WebJournalPapers/RiskAssessentSPS.pdf

[2]     Ancillary Services Presentation. Available: http://www.nyiso.com/public/webdocs/services/market_training/workshops_courses/nymoc/Ancillary_Services.pdf

[3]     J. K. Earle, “Functional unbundling of special protection systems as a required interconnected operating service in a deregulated environment,” MSEE Thesis, University of New Brunswick, 1997. Available: dspace.hil.unb.ca:8080/handle/1882/42522

[4]     H. Singh, “Introduction to Game Theory and Its Application in Electric Power Markets,” IEEE Computer Applications in Power, IEEE Computer Application in Power, vol.12, no.2, pp. 18-20, 22, Oct.1999.

[5]     J. Mepokee, D. Enke, B. Chowdhury, “Cost allocation for transmission investment using agent-based game theory,” International Conference on Probabilistic Methods Applied to Power Systems, Iowa State University, Ames, Iowa, September 12-16, 2004.

[6]     L. Fangxing, B. Rui, "Small Test Systems for Power System Economic Studies," Proceedings of the 2010 IEEE PES General Meeting, Minneapolis, MN, July 25-29, 2010.

Kundur Two-Area Test System

Kundur's two-area test system, from Prabha Kundur's book "Power System Stability and Control", is a power system utilized mostly for testing dynamics of solving stability issues.  Most researchers and engineers worked on this system to analyzed HVDC and FACTS impact on the transient stability. Other works were focused on small signal stability effect of such devices and/or Power System Stabilizers (PSS).

Single-line diagram of the Kundur two-area system.

The left part of the system is Area 1 and right part is Area 2. The ties (lines between buses 7, 8 and 9) are 110 km long thus the interconnection is rather not strong.

The system has dynamic data for the four machines and their exciters and stabilizers, aside from the power flow data. I took the power flow and dynamics data from "Implementation of an Adaptive Controller for Controlled Series Compensators in PSS/E". I adjusted the tie lines' data since the paper indicates that the lines were 150 km.

I will probably post power system dynamics tests using this system in the upcoming months (impact of PSS, critical clearing times, etc.).

You can download the test system built on PowerWorld v16 using this link.

Visayas Grid Approximate Model

I have developed the Visayas Grid approximate model using available public data following the procedure given in my approximate Luzon network model. For the Visayas model, I used the line parameters given in the book “Probabilistic Transmission Planning” as shown below.

The single-line diagram was posted previously at wesm.ph but not currently. Still, publicly, Nick Nichol’s website has it – link.

Again, this is approximate and does not attempt to replicate what NGCP or WESM is using. Nevertheless, this model can be utilized by electrical engineering instructors in teaching power systems, analysis and issues in the electric power industry. It can be useful for computer based laboratory exercises in power systems, transmission and distribution. It is useful for research for technical reports or thesis during senior year. If there are Filipino electrical engineering instructors open to discussing how to use this model, I am very willing to cooperate and we can do this via skype or google+.

Some notes on developing the Visayas grid model:

• The Visayas submarine cables are an important components together with the shunt reactors. Precise modeling of these components is not attempted.
• The nature of Visayas grid is that it is not a meshed power system but is a radial power system: from Leyte-Samar to Bohol and Cebu to Negros to Panay Island. This means power flow solution algorithm being used can be tricky unlike in meshed power systems like that of Luzon.
• Validating this model can be cumbersome, apply generation MW/cost bid parameters from wesm.ph and see if the locational marginal prices of this model and that of posted at wesm.ph matches up. If you would like to cooperate on this, I am open to doing it in PowerWorld. 

I’m not connected to PowerWorld, nor I’m endorsing it. It’s just that I am using it and because PowerWorld has practical power system applications like power flow, contingency analysis, shift factor calculations, optimal power flow, security constrained optimal power flow, short circuit analysis, and transient stability just to name a few. 

If you have questions or want the model, drop me an email at ebcano@gmail.com. Or download the model here.

Leyte Mindanao Interconnection Project

When the Leyte Luzon HVDC project was conceived, part of the plan was to extend the interconnection to Mindanao. From ABB, the figure below presents the interconnection projects.

The National Grid Corp. of the Philippines (NGCP) is now seeking approval for the LMIP, as reported by the Inquirer.

The Leyte Luzon link is a Current Sourced Converter (CSC) HVDC. This is the reason why reactive compensation is needed in its operation to support real power transfer and limitation of harmonics. When transmission lines are toppled in the Bicol region due to typhoons, the MW dispatch of the HVDC was curtailed due to the low short circuit capacity (SCC) in the area. Low SCC may result to commutation failure in the valves when voltage is weak on the AC system. This link also requires to operate between a minimum and a maximum MW bandwidth.

With the LMIP, I am guessing the technology of Voltage Sourced Converters (VSC) HVDC will be used. The side of Mindanao has been suffering with generation deficit thus will have low SCC and less reactive power support. VSC HVDC can be used in such case. Reactive power compensation is not required in VSC HVDC to transfer real power since it can operate independent of real and reactive power. Also, at high switching frequency for the VSC HVDC, less filtering or switchable harmonic filters are not needed. With this, if LMIP uses VSC HVDC, it can be utilized to start-up the Mindanao grid after a major outage or blackout since it can operate at low SCC or weak AC system. Another advantage is VSC HVDC can operate between 0% and 100% of rated active power in controllable in both directions without the need of DC voltage reversal since both Mindanao and Visayas need generation capacity to satisfy growth demand.

The Philippine power grids are one of the most advanced power systems in the South East Asian region having CSC HVDC, wind power and solar power integrated into the electric system. With the proposal of LMIP, new technology like VSC HVDC may find a way to connect and power the islands.

Using the Approximate Luzon Network Model for Power Engineering Education and Training

Educators and trainers can utilize the model in lectures or laboratories for discussion of the following:

1.    Power flow analysis
2.    Application of grid code limits on branch thermal capacity and bus voltages with or without outages
3.    Application of N-1 contingency
4.    Determination of maximum generation of an area with or without N-1 contingency
5.    Determination of  the limiting contingency for dispatching maximum generation of a plant
6.    Determination of the maximum generation that can be interconnected to a specific bus without violating grid code limits on branch thermal capacity and bus voltages with or without N-1 contingency
7.    Determination of   how much load growth can be accommodated without transmission/generation expansion
8.    Determination of   the generation margin/reserves at peak and off peak conditions
9.    Impact of enabling on load transformer taps on bus voltages
10.    Impact of limited reactive power capacity of a certain plant on bus voltages
11.    Impact of outage(s) of 500kV line(s) on the system
12.    Impact of outage(s) of 500kV transformer(s) on the system
13.    Impact of load power factor of the system or of an area on the system performance
14.    Impact of power contract transactions on the system performance applying grid code limits
15.    Application of load forecast for Luzon in the coming years and determine needed generation and transmission expansion

There might be other applicable analysis depending on the capability of the software being used, in this case Powerworld.  Thus, the list above is not exhaustive.

What Happened Here?

Historically, Metro Manila carries a whopping 50 to 60 percent of the total load of the Luzon Grid in the Philippines. Five bulk power substations deliver power to the metropolitan, if a disruption in the transmission lines connecting the load center to any of these substations, a power blackout is imminent in the area. And when your demand is cut into about half, the tendency is for system frequency and bus voltages to swing throughout the system.

When the crane cut off the transmission line, this produces an open circuit that may divert power flows in the power system and may produce a high impedance as seen by protection devices set to guard against system faults. Normally, the protection devices act faster for short circuit and slower to open circuits, since open circuit produces less fault (overload) currents.

Static Voltage Stability Analysis for Electric Subtransmission Systems

The study of voltage stability as indicated in the Philippine Grid Code (PGC) is the center of this paper. Standards and industry practice for voltage stability problem-solving are cited and were referred when provided an example simulation. Voltage stability is a must when looking at a power system if it can handle load growth and at the same time maintaining acceptable voltage levels at all system nodes pre and post-contingency. The static voltage stability simulations utilized practical solutions for voltage instability which are discussed and evaluated using Power-Voltage (PV) and Voltage-MVAR (VQ) curves. This report serves as a tutorial for practicing engineers on the important topic of voltage stability.

See paper by clicking this - Static Voltage Stability Analysis for Electric Subtransmission Systems

Voltage Security in Philippine Power Systems

Voltage security in power system planning and operations is an important factor of grid and distribution codes in the Philippines. The power system, system components and customer devices respond to the magnitude of voltage in a manner that would impact its operation. Power system interruption cascades when voltage collapse is experienced. Generators and other system elements tend to be saturated when keeping the voltage secured at certain limits. Customer devices might misoperate or may expel from its connection point when voltage is not appropriately within the devices’ operating regions. These are some of the reasons why it is important for a power system to be planned and operated in terms of voltage security. With areas where generation deficit is a problem, voltage security is hard to achieve since generators provide and assist in keeping the voltage secured in terms of reactive power management.

Philippine transmission systems must be operated in normal conditions within voltages of 0.95 per-unit up to 1.05 per-unit while distribution systems are expected to have voltage levels from 0.90 per-unit to 1.10 per-unit. These voltage magnitudes should be kept in steady-state conditions. Scheduled maintenance of lines and other components must be conducted making sure that these voltage performance standards must be kept. When a fault or a transient event occurs, the voltage levels can not deviate from 0.90 per-unit to 1.10 per-unit as long as 0.00833 seconds to 60 seconds. In this case, voltages magnitudes are classified as voltage sags or swells. Above 60 seconds, and the voltage levels are still deviating from the said limitations, these voltages are called long duration voltage variations.

Voltage security can be planned and operated by reactive power management and predicting the voltage profile in contingency analysis. Reactive power supervision entails updated reactive power capability curves of existing generators and list of static and dynamic reactive power devices installed in the power systems. When running a post-contingency voltage assessment, the reactive power capabilities of generators, reactors, capacitors and static VAR devices (SVDs), and operating points of tap-changing transformers, must be considered since these options can be enable in running appropriate post-contingency power flow solution. The area/zone/buses in study must be identified by the power system engineer(s) for proper scenario and conditions setting. The voltage ranges limits and voltage deviation limits must be defined for the area or zone or buses in study for analysis purposes. Together with the identification of the area in study, possible system contingency scenarios, assigning of voltage limits at the area in study the power flow solution or transient simulation can be automatically configured to report all the violations as a result of each probable contingency. Usually, in power flow computations, these can be presented in tables or graphs while for transient simulations, whether in phase domain or time domain, these voltages are analyzed using time versus voltage plots.

Philippine power systems just like any other power systems must be operated in voltage security. Voltage must be operated and planned accordingly for the power system and its components to operate securely. Codes define limitations and thus appropriate prediction and prevention of unsecured voltage can be analyzed. Knowledge of existing reactive power devices in the power system and proper coordination of these devices will enhance voltage security together with applying computer solutions that can automatically provide flexible programming of studying areas of voltage violations.