Published by Mike Falter, EE Power – Market Insights: Making Sense of Electric Vehicle Charging Options, September 07, 2022.
When it comes to charging your electric vehicle there are several options to consider based on speed, convenience and cost.
In the recent article on the Wallbox acquisition of COIL, EE Power briefly reviewed the Wallbox portfolio of EV charging solutions. EV charging solutions are not all the same and within just the Wallbox portfolio there are multiple solutions with varying charge rates, price points and installation requirements. Charging platforms can be suited to long or short-distance driving, public or home use and can have vastly different installation and operational costs.
In this article, EE Power takes a closer look at the different types of EV charging stations, their capabilities and use cases.
EV Charging Defined by Standards
To start, EV charging infrastructure is largely defined by two main standards. The J1772 SAE standard covers the general physical, electrical, functional and performance requirements to facilitate conductive charging of EVs in North America while IEC 62196 governs similar standards in Europe.
Fast DC charging solution. Image used courtesy of Wallbox
EV Charging Levels
EV charging can be categorized into three different Levels with different power outputs and charge rates:
Level 1 Charging – In North America and Japan Level 1 uses single phase 120 V AC, accessible from a standard residential 3 prong wall socket, to charge the EV through an EVSE (Electric Vehicle Service Equipment) cable and Type 1 J1722 plug that connects to the EV charge port. Level 1 chargers can deliver up to 20 A (2-3 kW @ 120 V) of charge power. This translates to about 5 miles of vehicle range per hour of charge time for a typical EV with 2.5 miles of range per kWh of battery capacity. At these slower rates, Level 1 systems are best used for overnight home charging or shorter-range driving. Note, there is no Level 1 charging in Europe since the standard residential voltage level is 240 V.
Level 2 Charging – These chargers use 240 V AC and are commonly used in public space applications like parking lots, although they can be used in home applications with the proper electrical infrastructure. The Wallbox Pulsar Plus is an example of a Level 2 charging solution. These chargers can be hardwired or connected through a properly rated NEMA plug to a 240 V wall socket. In North America, charge rates with Level 2 systems generally get to about 9.6 kW (40 A @ 240 V single phase) which translates to about 20-25 miles per hour of charge or 4-5 times the rate of Level 1 systems. In Europe, the Middle East and Africa Level 2 chargers use 3 phase AC that can deliver even more power, up to 22 kW (32 A, 3 phase) for residential applications.
Level 3 Charging – Also known as Fast DC charging, this is the fastest way to charge an EV with charge rates well above 100kW. Supernova from Wallbox is a fast-charging solution rated to 130kW that can add up to 120 miles of range to a typical EV in 15 minutes. The Hypernova model (planned for release in 2023) will be a fast charger capable of delivering 350 kW. At 350 kW the GMC Hummer EV pick up with 213 kWh battery capacity can be fully charged in just over 30 minutes. Level 3 achieves these high-power levels by charging at over 480 V DC.
EV Charging Levels. Image used courtesy of Central Hudson
Connecting to the EV Charge Port, Variations by Region
The plug, or connector, is what connects the charge station to the EV, allowing it to deliver power to the high voltage (400 V or more) lithium-ion battery powertrain. The Type 1 J1772 is the standard plug used in single-phase AC Level 1 and Level 2 charging applications in North America and Japan, and supports charge rates up to 10 kW. In Europe, IEC 62196 Type 2 plugs are used for 3 phase AC Level 2 charging applications up to 22 kW.
For DC fast charging the CHAdeMO plug configuration accommodates up to 100 kW and is most common in Japanese models. The CSS (Combined Charging System), or Combo plug, is a clever modification to the J1772 Type 1 plug in North America, or the IEC Type 2 plug in Europe, that adds two extra power contacts to the standard AC connector to support the higher DC charge rates. Due to its versatility, the CSS plug is quickly emerging as the standard for vehicles in North America and Europe. Telsa uses a proprietary plug for its network of fast charging stations but offers adaptors for use with public stations that use either CHAdeMO or CSS.
EV Charging Plug Type by region. Image used courtesy of Blink Charging
The EV charging port will often be configured to accept both AC and DC charging connectors and may have multiple ports. CSS sockets can support both Level 1/Level 2 AC and CSS “Combo” fast DC plugs.
EV charge socket configurations. Image used courtesy of the Driven
EV Onboard Charging Module and DC Fast Charging
Most EVs have an AC/DC converter on board that converts external AC power (Level 1 and 2 charging) to DC for charging the battery powertrain. However, in the case of fast DC charging, utility power is converted to DC externally, so the EVs internal AC/DC converter is bypassed allowing the batteries to be charged directly. In this manner, much higher rates can be achieved since the onboard AC/DC conversion typically limits rates in Level 1 and Level 2 applications.
DC chargers require a lot of power from the grid which can make the costs of operation and installation a lot higher compared with Level 1 or 2 systems. But the benefit is significantly faster charging times and a solution better suited to long range use cases like the US Interstate Highway System. $5 billion in funding was recently allocated through the federal Bipartisan Infrastructure Law to help expand the network of fast charging stations across the Interstate Highway System.
EV onboard charging module for AC and Fast DC stations. Image used courtesy of Springer
Architecture of a DC Fast Charging Station
At its heart, a DC fast charger is a power inverter that converts three phase utility power to the DC current needed to replenish the EV battery powertrain. Typically, line AC power is converted to DC through a controlled rectification stage using an IGBT bridge or similar circuitry. This can be followed by a suitable transformer isolated DC-DC conversion stage that conditions DC power specific to the needs of EV fast charging. Finally, the fast charger communicates with the onboard Battery Management System to properly monitor and regulate the flow of DC power to the EV battery pack.
Fast DC charger architecture. Image used courtesy of Infineon
Author: Mike Falter is Principal and Founder at TechLaunch Strategies.
Published by Prof. Silviu Darie, PhD (EE), Technical University Cluj Napoca, Romania, Honorary Member of the Romanian Technical Sciences Academy. Email: silviu.darie@enm.utcluj.ro.
Abstract: Based on the author’s experience one describes in field tested methodologies which can be employed in solving the frequency-based power quality problems. A generic approach is presented and an experience-based assessment is provided. After defining the main harmonics indices a case study with three (3) scenarios/cases with personal computer application is presented. PTW/SKM professional software is employed.
Keywords: Nonlinear loads, bus voltage total harmonic distortion, (THD-V), branch current total harmonic distortion, (THD-I), total harmonic demand distortion, (TDD), computer aided harmonic analysis, case study computer assisted.
1. Introduction
This paper provides the procedures that one should employ while Harmonics Analysis Investigation is of concern. One should highlight that Harmonic Analysis comprises several steps. The first step is to determine if harmonics exist in the given power systems. Usually this is given by a site survey and site measurements. Please notice that a close cooperation with the plant engineer is a “must-use” dialog. Once that the power system one line is provided and the filed data has been collected building the electrical model follows. Several professional power engineering software can be employed, such as PTW/SKM, ETAP, DigSILENT, CYME, EasyPower, Paladin DesignBase, etc. While the power system is built, all the power system components and system layout should in detail be considered. Once the model is completed, the harmonic sources are generated and then injected at the existing points of harmonics sources. The power system model has to be consistent with the requirements of IEEE 519-2014, IEEE-399 (Brown Book) and IEC 500 standards. As a rule, once the power system is built in any professional power software, always run Power Flow. The convergence of Power Flow demonstrates that the system is feasible and the input data is consistent.
1.1 Methodology
Based on the site survey the study power system layout, power system components data and system operation scenarios are collected. In general, the Harmonics Analysis study is conducted to determine:
• Bus voltage waveform, bus voltage spectrum and bus voltage total distortion THD-V % at Point of Common Coupling (PCC) or at all of the power system panels in the system model, if the project owner requires; Branch current waveform, branch current spectrum and branch current total Distortion THD-I % at all of the branches which fed the panels in the system model;
• One recommends computing the total “Harmonic Voltage” and current indices; this gives the bus voltage fundamental magnitude V1 and the bus actual voltage Vrms considering harmonic contribution; branch current fundamental I1 and branch actual Irms current considering the harmonic contribution;
• The harmonic voltage and current.
1.2 Assumptions
When necessary, assumptions are made based on mutual agreement and with suitable technical and/or the other relevant applicable arguments. Bus IDs, branch names and equipment characteristics are shown on the Computer Program model drawing and project database. Equipment Names should be input as per “project as built”. However, in completing the plant model one uses plant one-line drawings and data from site survey of the existing equipment.
2. Harmonics investigation
2.1 Harmonics indicators
There are several indexes which can be used to measure the harmonics impacts to the power network. The most commonly used indexes are, [3, 6, 8, 9, 11]:
• Total harmonic distortion; • Total harmonic distortion for bus voltage, V-THD (%); • Total harmonic distortion for current, I-THD (%); • Total harmonic demand distortion, TDD for branches; • Telephone interference factors, if this is requested IT.
2.2 Harmonics voltage distortion
Harmonic voltage distortion index Vn% is the r.m.s. amplitude of a harmonic voltage of order “n” expressed as a percentage of the r.m.s. amplitude of the fundamental, [8, 11]:
.
with, n ≠ 1
2.3 Harmonics current distortion
Harmonic current distortion In% is the r.m.s. amplitude of a harmonic current of order “n” expressed as a percentage of the r.m.s. amplitude of the fundamental, [8, 11]:
.
with, n ≠ 1
2.4 Total harmonic bus voltage distortion, V-THD, %
Total harmonic bus voltage distortion, (THD-V)% is defined as, [8]:
.
where: V2, V3, …Vn are the individual harmonic voltage magnitudes, in V; V1 is the fundamental frequency of the voltage magnitude, in V.
The total harmonic distortion factor, THD can be computed by using the field measurements results, or by using a professional software such as HI_WAVE, of PTW/SKM – [www.skm.com] to simulate the network and compute THD at different bus location. By setting the program, THD is calculated for all network busses and any THD larger than a target value is flagged in the output results
Bus total harmonic distortion, (THD-V)% should normally be less than 5%. The IEEE 519-2014 Standard recommends, [3]:
Table 2.1 Bus total harmonic distortion, [3]
.
Note: One should note that the voltage distortion level is dependent on the system impedance characteristics and the harmonic current injected by the individual harmonic sources.
The table 2.2, below, gives the IEEE 519 -2014 current limits as function of the short circuit factor Isc / ILoad,[3].
Table 2.2 Harmonic current limits, [3]
.
where: Isc is the short circuit current at the PCC, in A; ILoad the load current, in A at the PCC.
3. Computer assisted harmonics analysis
Based on the author’s experience one proposes the procedure listed in in Figure 4.2. Please note that once the system is modelled, power flow always should be run. The convergence of the power flow demonstrates that the modeled system is feasible and the input system data is consistent. The guidelines that address harmonics share a common goal: to maintain the quality of electrical power at the point of common coupling (PCC). One meets such a goal if one limits the harmonics-induced voltage distortion at the point of common coupling, PCC. The “point of common coupling” is the interface between sources and loads.
Most experts select the primary of a feeder transformer as the PCC and the distortion is measured here.
3.1 Point of common coupling
On recommends selecting the PCC on the primary side of the main distribution power transformer for industrial projects, or the secondary side for commercial supply, Figure 3.2, [3].
Figure 3.2 PCC Location as per IEEE 519-2014 Standard, [3].
3.2 Harmonics investigation procedure flow chart
Based on a large number of projects completed by the author one provides a practical guide on harmonics analysis and harmonics investigations, Figure 3.5.
In harmonics-based power quality investigation, the consultant engineer should well understand the study power system and should recommend a site visit. However, based on the site survey the consultant engineer will collect all the power system data, network structure, electrical loads, type of cabling and earthing, machine parameters as built. Also, the operating regimes will be considered. Site visits are the best time to compare, update and correlate the project data with the real data collected from the field.
Harmonic measurements may be made on power systems at the consultant engineer request for several reasons, such as:
• To identify the source of harmonics and harmonic source location; • To determine the spectrum of harmonics current and voltage; • To confirm the findings of a simulation; • To verify previous harmonic studies (if this exist) and check the filter design impact; • To perform measurement of harmonic mitigation devices.
During the site survey the network database is collected and saved. The following data is needed:
• Network one-line diagrams as built; • Network parameters including feeder longitudinal and shunt impedance, power transformer name plate data and parameters, reactor name plate data, motor data and operating regime, generator data and running regime; • Electrical loads and generation; • Load profile and structure; • Type of harmonic source and harmonics components.
Note: Computer programs are used to calculate the levels of harmonic distortion, harmonic current flows and the effects of different filter designs;
The database should be organized consistent with the requirements of the software to be employed for network investigation.
4. Computer aided power system harmonic analysis
4.1 Scope of study
The Harmonic Study is to compute the Total Demand Distortion (TDD) and Bus Total Harmonic Distortion (THD-V) at the defined PCC. The purpose of imposing limits on the harmonics emissions is to ensure that the current and voltage distortions at the Point of Common Coupling (PCC) be kept at a low level as possible. Thus, the other customers connected at the same point are not disturbed.
The Harmonics analysis may be completed by several professional software program, and should be consistent with the requirements of the current standards (IEEE 519-2014 Standard, IEEE Brown Book, IEC 509, etc.). In this paper the PTW/SKM industrial power software is employed, www.skm.com, [13].
4.2 Methodology
While generating the power system model, the power system data is provided by the electrical contractor and the design team, the joint venture personnel and the design office of company that is involved in the study. Also, one needs to get the data obtained from the vendor’s submittals on the equipment that is being installed in the project.
The harmonics analysis is performed to calculate the followings:
• Bus voltage waveform, spectrum and bus voltage total distortion. The bus THD-V at all the panels in the power system model; • Branch current waveform, branch current spectrum and branch current total distortion, the current THD-I; • The total harmonic voltage and current indices; these give the bus voltage fundamental magnitude V1 and the bus actual voltage Vrms considering harmonic contribution; branch current fundamental I1 and branch actual current Irms considering the harmonic contribution; • Compute the harmonic voltage and current; • Design filters if needed.
The Harmonic study should include:
• A detailed computer model of the electrical power system using the requested professional software program. The model should allow any system studies, “what if” scenarios, system performance analyses, power quality investigation, etc.; • Input data and assumed data reports; • Harmonic Analysis.
4.3 Computer aided harmonic analysis
The power system harmonics analyses are performed to assess the followings:
• Frequency Scan (Impedance versus frequency analysis). It is important analysis for predicting the system resonances in the distribution system. Peaks of the impedance plot indicates parallel resonance conditions while valleys are an indication of series resonance;
• Voltage Distortion Analysis – voltage harmonic distortions should be calculated at all of the buses in the distribution system and the results will be compared with IEEE-519 harmonic limits;
• Current Distortion Analysis – current harmonic distortions should be calculated for all of the distribution system and the results will be compared with IEEE-519 harmonic limits.
Figure 4.1 Harmonic Analysis Flow Chart
The Figure 4.1 provides the recommended procedure and steps for harmonic analysis:
1. Generate the power system model as built;
2. Obtain from the company the relevant data and requirements at the point of common coupling. These must include: minimum and maximum fault levels for different system conditions; permissible limits on harmonics including distortion factors and IT factor; The criteria and limits vary considerably from country to country;
3. Complete harmonic analysis for the base system configuration;
4. Compute harmonic voltage distortion factors and IT (if this is requested) value at the point of common coupling;
5. Examine the power system results: bus voltage drops, feeders and power transformers loading, power factor, power system losses;
6. Go back to step 1 or step 4, depending on whether the network data or only the parameters of the analysis need to be modified;
7. Check the power system losses without and with the harmonics present in the system;
8. Calculate the requested shunt capacitor ratings to reach the desired power factor; Apply a detuning reactor if a resonance condition is found. Go back to step 4;
9. Design and add harmonic filters if the harmonic distortion factors value at the point of common coupling exceeds the limit imposed by the utility.
5. Harmonic analysis: case study – generic power system
One considers the study system presented in the Figure 5.1. As can be seen, at the MAIN switcher several circuits are reticulated; a static load L3, one M1 motor circuit and a 400 V PNL-1; 2 motors M2 and M3 are connected to the panel PNL-1; motor M2 is supplied via a VFD system; the total static load connected at the PNL-1 is represented by a composite static load L2 which has harmonics generated by the fluorescent system.
The harmonic analysis is performed by employing PTW/SKM (www.skm.com) professional software. The PTW/SKM is a powerful industrial power software for designing and analysis of power system. It is worldwide employed by consultant engineers, designers and utility engineers. The PTW/SKM is on the market for more than 38 years. It has a powerful Graphical User Interface (GUI) with several power system calculations, a large power system database and intuitive display information. The PTW/SKM is used by over 35,000 engineers worldwide, offering specialized Power Tools design and powerful modeling and documentation capabilities. Professional trainings are provided for PTW/SKM users.
Figure 5.1 a) Generic Power System Layout; Model View
Figure 5.1 b) Detailed Model View at PCC area
For this study, three (3) Scenarios/Cases are considered as flows:
Scenario 1 /Case1: all harmonic sources are ON, harmonic filter is OFF and the power factor shunt capacitor CAP-1 is OFF;
Scenario 2/Case2: normal, with shunt capacitor CAP-1 On, all harmonic sources ON, harmonic filter OFF and the power factor shunt capacitor CAP-1 is ON;
Scenario 3/Case3: the power factor shunt capacitor CAP-1 is OFF, all harmonic’s sources are ON, the harmonic filter is ON.
5.1 The power system model input data
A summary of the power system model input data is listed in the Table 5.1.
Table 5.1 Power System Model Input Data
.
5.2 The harmonic sources
The harmonics sources present in the system study are:
• Typical 6 Pulse IGBT at the motor M1; • The fluorescent lights at the static load L1.
The Figure 5.2 and the Table 5.1 provides the harmonic sources characteristics:
Figure 5.2 a) 6 Pulse IGBT; b) Fluorescent lights
Table 5.2 Loads and Motors with Harmonic Source Models
.
6. Harmonics investigation
6.1 Frequency scan
Frequency scan is performed with the shunt capacitor bank OFF, Case 1 and with the shunt capacitor bank ON, Case 2, Figure 6.1.
Figure 6.1 Frequency Scan: The capacitor bank CAP-1 is OFF, red color; The capacitor bank CAP-1 is ON, blue color
Note: As can be seen from Figure 6.1 the shunt capacitor CAP-1 generates resonances and it represents an harmonic amplifier.
The frequency response shape at a given bus depends on the existing or not of a shunt capacitor bank in the power system. Several problems may occur when the system response exhibits a parallel resonance near one of the harmonic components which are in the system (usually the 5th or the 7th harmonic). The resistive load provides damping near these resonant frequencies. The combination of these two factors determines whether or not a harmonic problem will exist at a particular bus.
6.2 Bus voltage waveforms and distortion spectrum
One investigates both the voltage distortion and the distortion spectrum at the MAIN bus, above the PNL-1 (PCC) bus in order to monitor the harmonic penetration towards Utility source, Figure 6.2
There are several indexes which can be used to measure the harmonics impacts to the power network. The most commonly computed indexes are:
• Bus total harmonic distortion, V-THD (%); • Branch current total harmonic distortion, I-THD (%); • Total harmonic current demand distortion, TDD for branches; • Telephone interference factors, IT.
Figure 6.2 Bus M1 and bus PCC – Voltage Waveforms and Spectrum, Case 1
In Figure 6.2 a) represents the MAIN bus voltage waveforms, in blue color and the bus PNL-1 (PCC) in red color.
Figure 6.2 b) represents the bus voltage distortion spectrum.
The total voltage distortion, for Scenario/Case 1 (CAP-1 OFF) is listed in the Table 6.1. One should note that the voltage distortion level is dependent on the system impedance characteristics and the harmonic current injected by the individual harmonic sources.
Table 6.1 Total Bus Voltage Distortion, Case1
.
While the shunt capacitor is ON, Scenario/Case 2 which is the normal regime, large bus voltage distortion is induced at the PNL-1 (PCC) bus, see Figure 6.3 in the blue color. One can see that the 5th and the 7th harmonic orders are generated.
Figure 6.3 PNL-1 (PCC) – Bus Voltage Waveforms and Spectrum, Scenario/Case 2, Normal Regime
The shunt capacitor bank works as an harmonic amplifier; the curves in blue color. The 5th and the 7th harmonics are increased to 4.788 and 6.759 respectively, in per unit, Figure 6.3.
While the shunt capacitor is ON, Case 2, the bus voltage distortion is increased, see Table 6.2.
Table 6.2 Total Voltage Distortion, Case 2 the Capacitor Bank CAP-1 is ON
.
Harmonic studies are performed to determine the harmonic distortion levels and filtering requirements within a study facility. In general, field measurements and computer simulations are used to characterize the impact of the nonlinear loads. Based on the computer simulations the required harmonic filter is selected and sized. The application of harmonic filters will significantly alter the frequency response of the power system.
The level of the harmonic distortion is dependent on two important factors:
• The level of harmonic distortion; harmonic currents are generated by loads which have nonlinear “voltage-current” characteristics; • The number and sizes of the nonlinear devices at a given bus determines the level of harmonic current generation.
If local resonances exist, then it is also possible the harmonic problems to occur at buses remote from the harmonic sources. If capacitors are applied at any locations, the potential for resonance problems must be considered carefully.
The following technical solutions can be employed:
• Harmonic filtering; • Installing capacitors banks and filters; • Increase pulse number of electronic devices; • The use of custom power technology and products.
For harmonic mitigation one uses the tuned harmonic filters. Harmonic filtering is a typical mitigation technique which is employed by the utility, industrial and commercial systems. Basically, there are two types of filters:
• Passive filters, where the filter components are passive elements such as resistor, inductor and capacitor; • Active filters, where the filter has a controlled current or voltage source.
7. Harmonic filter design
For harmonic filter design on should start by investigating the actual power factor. For this, one runs the power flow and check the system power flow results. The convergence of the power flow demonstrates that the system is feasible and the input data is consistent. The bus voltages and the branch current should be within the standard limits.
Further, one may consider the followings rules of thumb in designing the harmonic filter, [11, 13]:
• Always start mitigating the harmonics with the lowest harmonic order; • Connect the filter near to the harmonic sources; • It is a good practice to tune the filter to (3 to 5) % below the harmonic order to be mitigated; • The filter reactive component C (Q) will compensate the power factor close to the desired power factor (client request); • Increase the filter Q to produce higher harmonics mitigation: in this particular situation this is dangerous due to the excessive increase of power factor at the panel where filter is installed. During a plant operation with low system loading, this issue will generate problems with the system reactive stability; • Above tuned frequency the harmonics are absorbed; • Below tuned frequency the harmonics may be amplified.
7.1 Power flow results
While shunt capacitor bank is OFF, the power factor at PNL-1 (PCC) is LFPF = 0.87. Let’s assume a desired power factor of 0.97; in this case, a 200 kVAR reactive power is requested. This amount of reactive power will be considered while harmonic filter will be sized.
Figure 7.1 Power Factor Improvement, [13]
Based on power flow analysis, the comparative results are provided by the “SKM Visualizer” as listed in Table 7.1.
The power factor is computed at the Power Factor Measurement Cubicle for each Scenario/Case. One has to note that the power flow analysis application does not consider the harmonics.
Table 7.1 Comparative Power Factor Results: CAP-1 OFF, PF = 0.87; CAP-1 On, PF = 0.99
.
7.2 Harmonic filter sizing
To reach the requested power Factor of 0.97 the harmonic filter capacitor should be 250 kVAR. With this capacitor bank value, the harmonic filter sizing becomes:
Figure 7.2 Harmonic Filter Sizing
Single Tuned Filter Data: FLTR-1 Rated voltage 400 V; Capacitor size: 250 kVAR; Harmonic order to tune 5.00; Q Factor: 80; C = 4973.60009.
Once the harmonic filter has been sized and connected to PNL-1(PCC) bus, one performs harmonic analysis by employing the PTW/SKM professional software. The harmonic analysis results are provided via the Tables 7.2 -7.20. Some of the tables are listed in the body of the paper, the others are listed in the Appendices which are part of this paper. The following harmonic analysis were completed for Scenario/Case 3 FLTR-1 ON:
• The comparative harmonic studies results: Total Bus Harmonic Distortion V(THD) % for each Scenario; Table 7.2 for the power system components; • Voltage distortion summary, Table 7.3; • Total voltage distortion, Table 7.4; • Total Current Distortion, Table 7.5; • Current Distortion Summary, Table 7.6; • Harmonic Voltage Spectrum Report, Tables 7.7 to 7.12; • Harmonic Current Spectrum Report, Tables 7.13 to 7.16; • Harmonic Filter Design, Harmonic Filter Data, Table 7.17; • Filter Spectrum Report, Table 7.18; • Harmonic Current Spectrum Report, Table 7.19; • Power system losses, Table 7.20.
Table 7.2 Comparative harmonic studies results: Total Bus Harmonic Distortion V(THD) %; FLTR-1 OFF versus FLTR-1 ON
.
7.3 Bus voltage waveforms and spectrum
With harmonic filter ON, the bus voltage waveforms and spectrum are listed in the Figure 7.3.
Figure 7.3 Bus Voltage Waveforms and Spectrum a) CAP-1 ON FLTR-1 OFF; b) CAP-1 OFF FLTR-1 ON
One can be seen a large reduction of the bus total harmonic distortion at the PNL-1 (PCC) from 6.7 % to 3.5 %. Also, one can see the improvements of the system performances above the PNL-1 (PCC) bus towards to utility bus.
Table 7.3 Voltage Distortion Summary, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.4 Total Voltage Distortion Summary, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.5 Total Current Distortion Summary, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.6 Total Current Distortion Summary, Case 3; CAP-1 OFF, FLTR-1 ON
.
Table 7.7 Total Current Distortion Summary, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.8 Harmonic Voltage Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.9 Harmonic Voltage Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.10 Harmonic Voltage Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.11 Harmonic Voltage Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.12 PNL-1 (PCC) Harmonic Voltage Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON
.
Table 7.13 Harmonic Current Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.14 Harmonic Current Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.15 Harmonic Current Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.16 Harmonic Current Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.17 Passive Filter Data, Harmonic Current Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.18 Filter Spectrum Report, Case 3; CAP-1 OFF, FLTR-1 ON.
.
Table 7.19 Total System Power Losses, No Harmonics in the System, CAP-1 OFF
.
Table 7.20 Total System Power Losses, No Harmonics in the System, CAP-1 ON
.
Table 7.21 Total System Power Losses, with Harmonics, Scenario/Case 2: CAP-1 ON, FLTR-1 OFF
.
Table 7.22 Total System Power Losses, with Harmonics, Scenario/Case 3: CAP-1 OFF, FLTR-1 ON
.
Note: Harmonics filters improves system power factor and reduce power losses, see Table 7.22
8. Conclusions
The paper provides an infield tested methodology and a guide in solving harmonics-based power quality computer assisted. The harmonic studies are performed by employing PTW/SKM professional software.
Harmonics generate problems if the power system is not designed to handle them. One notes that a large voltage distortion the PCC is acceptable as long as sensitive equipment is not affected. However, it is always important to consider the presence of harmonics and to try to minimize them by employing the appropriate low distortion electronic ballasts and reactors for PWM. Mitigating the harmonics will improve the power factor in the facility, and will also save energy by reducing power losses in power system components. Any time when there will be a considerable increase of non-linear loads, it is important to check the power system components loadings to prevent problems.
Based on the several projects completed by the author, the followings may be considered, [10, 11]:
• The harmonic limits are recommended for both voltages and currents.; • Both the system owners or the operators and the users must work cooperatively to keep actual voltage distortion below the standard levels; • The end-users should limit the harmonic current injections; • One needs to highlights that the recommended limits of the THD apply only at the PCC bus and should not be applied to either individual equipment or at locations within a user’s facility, [3, 8]; • Harmonic mitigation does not generate energy savings. Harmonics do not affect the real power absorbed by the loads; • K-factor is based on the THD-I present in the system; • K-factor does not mitigate harmonics. It permits the power transformer to operate in the presence of harmonics as long as transformer is not overheated; • Total Demand Distortion (TDD) compares the amplitude of the current with considered harmonics at any operating point to the demand (full) load fundamental current of the loads, [3]; • Total Demand Distortion (TDD) is based on the measured levels that define the harmonic current amplitudes and the maximum fundamental current of the loads; • One needs highlighting that the metering devices do not measure the TDD; • It’s important to note that the loads may cause some harmonics, while others harmonics may be from the power sources; • The only way to measure what is coming from a utility is to shut off all nonlinear loads in the plant and measure the voltage total harmonic distortion (THD (V)); Harmonic currents caused by the variable frequency drives (VFDs) depend on the impedance of the electrical system and the impedance in the VFD and/or the impedance in front of the VFD; • Typically, “Pulse Width Modulation”, or PWM with diode rectifiers does not have DC choke impedance and input line reactors, [6]; • The current practice recommends do not install line reactors in medium voltage system; If the harmonic current distortion is high, resonance or capacitor destruction caused by harmonic heating are possible; • If the study bus has greater than 50 % nonlinear loads, then active filters should be considered for both power factor correction and harmonic filtering purposes; • Harmonics filters improves system power factor and reduce power losses.
Contributions: The paper presents the author’s in-field tested methodologies for harmonics investigation, practical flow chart for harmonics investigation, large conclusion based on the author’s experience on harmonics investigation computer based; the step-by-step harmonic filter design and the impact of harmonics to distribution power systems
9. References
1. IEEE Standard 141-1993, IEEE Recommended Practices for Electric Power Distribution for Industrial Plants, (IEEE Red Book). 2. EN IEC 61000-3-2, European Standard, 2019. 3. IEEE 519-2014. https://www.elspec-ltd.com/understanding-the-ieee-519-2014-standard-for-harmonics/. 4. Cheng, J. IEEE Standard 519-2014. Compliances, Updates, Solutions and Case Studies. Schneider Electric. 5. Reducing of Harmonic Distortion. ASIAN Electricity, September 2003. 6. EPRI Electric Power Research Institute. Proceedings: Second International Conference on Power Quality. End-Use Applications and Perspectives. PQA’92. Volume 1&2, Albany, California, 1992. 7. George J. Wakileh Power Systems Harmonics. Fundamentals, Analysis and Filter Design. Springer, 2001, 506 pp. 8. Jos Arrillaga, Neville R. Watson Power Systems harmonics. Second Edition, John Willey & Sons, Ltd, 2003, 399 pp. 9. Silviu Darie, Harmonics Investigation; Part 1 Overview, Harmonics Indices, IEEE & IEC Standards. Rev. Energetica Nr. 10/2020, Volume 68, ISSN: 1553-2360. 10. Silviu Darie, Harmonics Investigation; Part 2 Computer Aided Harmonics Studies, Rev. Energetica Nr. 11/2020, Volume 68, ISSN: 1453-2360. 11. Darie, S., Computer Aided Harmonics Investigation (in English), Blue Print House Cluj, 2019, 170 pp. 12. Mekhamer, S.F., Abdelaziz, A.X., Ismael, S.M. Harmonic Analysis Studies Applied to Industrial Electrical Power Systems. ETASR-Engineering Technology & Applied Science Research. Vol. 3, No. 4, 2013. 13. *** Power Tools for Windows (PTW). HI_WAVE Reference Manual. Electrical Engineering Analysis Software for Windows. Copyright, 2009, SKM Systems Analysis, Inc., USA.
Author:Prof. Silviu Darie, PhD (EE)
Author: Prof. Silviu Darie, PhD (EE), Technical University Cluj Napoca, Honorary Member of Romanian Technical Sciences Academy, Former VP Power Analytics Corporation, USA.
Prof. Dr. Daries has more than 20 years’ work experience with Power Analytics products, and nearly 40 years of university-level electrical engineering instruction and industry consultancy in power system analysis computer applications, electrical power quality, transmission pricing, embedded generation, computer aided power system analysis and design. In addition to earning both his doctorate and master’s degrees in electrical engineering, he has authored or co-authored hundreds of technical books, student manuals, technical papers, and research projects.
Dr. Darie is a former professor of power systems and electrical engineering in Technical University of Cluj Napoca, Romania, and University of Cape Town, South Africa, as well as a former visiting professor in École polytechnique fédérale de Lausanne, Switzerland. He has received several awards and recognitions throughout his years of expertise including the Award Professor for Life of Faculty of Engineering, University of Cape Town 1993, Romanian National Research Award. Since 2005 he is the Vice President of Consulting and Engineering for Power Analytics Corporation.
Dr. Darie led nearly 180 electrical power projects worldwide; he constructed 18 prototypes designed for mass production, holds three patents, and is experienced in most leading software programs for electrical engineering. He has provided services to clients worldwide, and is a registered professional engineer in Romania, South Africa, and New Zealand.
Contact address: Prof. Silviu Darie, Ph.D., P.E., Romania: Bd. 21 Decembrie 1989, No. 104 Bl. L1, Sc. 1, Ap. 8 Cluj Napoca, 400124 Romania Mobile: +40728312222 Email: silviu.darie@gmail.com, Silviu.darie@enm.utcluj.ro
Published by Ahmad Rizal SULTAN1, Mohd Wazir bin MUSTAFA2, Makmur SAINI3, Ahmad GAFFAR4 1,3,4 Politeknik Negeri Ujung Pandang, South Sulawesi-Indonesia 2Universiti Teknologi Malaysia, Faculty of Electrical Engineering, Johor-Malaysia
Abstract. The aim of this paper is to detect the single line to ground fault on the unit generator- transformer. A new ground fault detection scheme based on the extraction of energy and statistical parameters from wavelet transform based neural network is proposed. The faulty current signals obtained from a simulation were decomposed through wavelet analysis into various approximations and details. The simulation of the unit generator-transformer was carried out using the Sim-PowerSystem Blockset of MATLAB. The energy and statistical parameters analysis involved measured of the dispersion factors (range and standard deviation) of wavelet coefficients. Regarding the ANN performance, the errors in the SLG fault detection of ANN were under 1 %. The results indicate that the proposed algorithm was accurate enough in differentiating a single line to ground fault and un-fault for a unit generator-transformer.
Streszczenie. Przestawiono metodę detekcji nieprawidłowości w uziemieniu jednostki generator-transformator. W nowej metodzie wykorzystano transformatę falkową I sieć neuronową. Symulację przeproprowadzno wykorzystując Sim-PowerSystem Blockset of MATLAB. Uzyskano błąd pomiaru poniżej 1%. Detekcja nieprawidłowości uziemienia w jednostce generator-transformator z wykorzystaniem transformaty falkowej i sieci neuronowej
Keywords: ground-fault detection, unit generator-transformer, wavelet transform, neural network Słowa kluczowe: nieprawidłowość uziemiania, jednostka generator-tranasformator, transformata falkowa
Introduction
Small current Ground-Fault (GF) detection has been a major concern in protective relaying for a long time. Relaying engineers and researchers often face the challenge of developing the most suitable technique that can detect faults with reasonable reliability to secure the run of a power system [1]. In general, a step up transformer at an electric power station can be categorized either as a unit generator-transformer configuration, a unit generator-transformer configuration with generator breaker, a cross-compound generator or a generator involving a unit transformer [2,3]. A GF on the transmission line or busbar can affect the system configuration of the generator.
Several methods have been reported for generator GF protection [4]. These methods have been developed based on conventional method, third harmonic method, sub-harmonic injection method and numerical protection method. Fault detection and classification algorithms based on Wavelet Transform (WT) and Artificial Neural Network (ANN) was proposed in [5, 6].
Various feature extraction methods based on WT have been used for the detection and classification of fault. Reference [6] describe fault location techniques in power system based on traveling wave using wavelet analysis and GPS timing. Fault classification algorithm based on energy and wavelet entropy in transmission have been proposed in [7, 8]. Reference [9-11] describe the feature extraction method based on fast WT, a fault index and wavelet power for use to detect the stator faults in the synchronous generator. Extraction of a statistical parameter as fault detection has been used for fault detection in previous studies, but only used standard deviation, kurtosis and skewness [12]. Meanwhile, the statistical feature parameters include kurtosis, skewness, crest factor, clearance factor, shape factor, impulse factor, variance, square root amplitude value and absolute mean amplitude value to fault diagnosis in rotating machine as described in reference [13]. The new approach as proposed in this paper includes energy and dispersion factor of statistical parameters on single-line to ground (SLG) fault detection.
The novel method for GF detection uses energy and dispersion factor of statistical parameters, which involve calculating the Energy, Range (R) and Standard Deviation (STD) values of wavelet coefficients, which are included the analysis in this paper. In the analysis, the GF signals were computed by using Discrete Wavelet Transform (DWT). The GF detection was carried out through the analysis of value of energy, R and STD of the current wavelet coefficients, including the detail and approximate of wavelet coefficients to distinguish SLG-fault.
Energy and Statistical Parameters Extraction Method
A WT is a powerful tool for feature extraction of the transient signals. WT has been applied in many literatures for feature extraction of transient fault signals. The differences among modifications of this method are: different types of mother wavelet, various numbers of decomposition level, and state of calculating the energy or entropy features. There are many types of mother wavelets such as Haar, Daubechies, Symlets, Meyer, Dmeyer, Morlet. The optimal choice of the mother wavelet plays a significant role for detection various types of transient signals. The optimum wavelet for extracting signal information is that can generate as many coefficient as possible to represent the characteristic of signals. In this paper, DWT was used for feature extraction, which provided high time and low frequency resolution for high frequency and high-frequency resolution and with low time resolution for low frequencies. The DWT was calculated by using the following equation [14]:
.
where “g(k)” is the mother wavelet, “x(k)” is the signal input and a,b are the scaling and translation parameters.
DWT was implemented by using high-pass filter and lowpass filter respectively [15], defined as:
.
where “yhigh(k)” is the output from the high-pass filter called Detail (D) and “ylow(n)” is the output from the low-pass filter called Approximation (A). For the 3-level decomposition, the original signal is split as shown in Figure 1. The original signal S is represented as A1 + D1, A2 + D2 + D1, A3 + D3 + D2 + D1.
The mean idea of making a feature extraction is to reduce the amount of information, either from the original waveform or from its transformation format. In this study, for feature extraction process, the coefficient features of wavelet such as wavelet energy, R and STD value of wavelet coefficient had to be calculated.
Fig.1. Decomposition tress of wavelet transforms
a. Wavelet Energy
The wavelet energy is the sum of square of detailed wavelet transform coefficient [16]. The energy of a wavelet coefficients is varying over different scales depending on the input signals. The wavelet energy of coefficient c(t) can be defined as follows:
.
with appropriate scaling coefficients aj for the coefficient cj obtained from the corresponding signal “s(t)”. The energy of signals is contained mostly in the approximation part and a little in the detail part [17]. For example, the approximation coefficient at the first-level contains more energy than the other coefficients at the same level of the decompositions tree. Because the faulty signals have high-frequency components, it is more distinctive to use energy of detail coefficients [18].
b. Dispersion Factor of Statistical Parameters
In descriptive statistics, the concept of range has a more complex meaning. The range is the size of the smallest interval which contains all the data and provides an indication of statistical dispersion. It is measured in the same units as the data. Since it only depends on two of the observations, it is most useful in representing the dispersion of small data sets [19].
STD is a number used to tell how measurements for a group are spread out from the average, or expected value. The STD of statistical parameters in wavelet detail coefficients are estimated from the equations:
.
where “x” is the data vector and the “n” the number of elements in that data vector. The STD of the output signals is the square root of the data vector variance. This feature provides information about the level of variation of the signal frequency distribution [20].
Artificial Neural Network Pattern Recognition
ANN is very good at pattern recognition problems. An ANN with enough elements can classify any data with arbitrary accuracy. They are particularly well suited for complex decision boundary problems over many variables [21]. The use of pattern recognition for power system security analysis was first investigated in 1968. Since ANN can fully be applied for pattern recognition, they have been widely investigated for transient classification [22]. The ANN can be used to solve power system protection problems, particularly those where traditional approaches have difficulty achieving the desired speed, accuracy, and selectivity [23-25]. Pattern recognition for partial discharge in GIS based on pulse coupled neural networks and wavelet packet decomposition have been proposed in [26].
In pattern recognition problem, a neural network can be used to classify input into a set of target categories. In this paper, a set input of energy wavelet coefficient and dispersion factor in statistical parameters of wavelet coefficients are used for input against set ground-fault or un-fault target categories.
Proposed Method
The block diagram of proposed SLG-fault detection algorithms is shown in Figure 2. The first step of the detection module was to get the current samples from Sim-PowerSystem Blockset of MATLAB simulation. The fault current signals were then computed by DWT. The fault detection was carried out through the analysis feature extraction of dispersion factor or the current energy of wavelet coefficients. Feature extraction of dispersion factor and the energy of the wavelet coefficients are analysed for comparison. The block diagram is explained in steps:
– Step 1: The fault current signals are obtained from a simplified power system model (Figure 3) for GF simulation using Matlab-Simulink.
– Step 2: DWT of the fault signals are obtained and analysis using MATLAB software.
– Step 3: The wavelet coefficients of the fault signals are obtained using signal decomposition.
– Step 4: The extraction of energy and dispersion factor of statistical parameters (R and STD value) of wavelet coefficients from WT in various fault simulations are fed to ANN and trained.
– Step 5: Energy and statistical parameters of WT based ANN distinguishes GF from normal condition.
Fig.2. Block diagram of a proposed algorithm
Fig.3. Simulated power system model for ground fault
A suitable unit generator-transformer model is required to characterize the different condition during SLG-fault. GF simulations were established using Sim-PowerSystem Blockset of MATLAB, where M-file MATLAB was used for GF detection. The simulated power system models for GF simulation are shown in Figure 3. The data of a generator (G1=G2) 25 kV with various generator grounding method, the transformer (transformer- 1=transformer-2) 25/150 kV with Yn-Yn transformer connections. Simulation was carried out at various fault locations includes primary and secondary side of a transformer-1, and at generator bus. Fault current was taken from the generator bus (Bus-1).
Analysis of Simulation Results
Designing SLG-fault detection on unit generator-transformer models follows a number of systemic procedures. In this paper, there are three basics steps:
(1) signals decomposition, (2) feature extraction and (3) ANN trained and verified.
(1). Signals Decomposition
In this paper, the energy and dispersion factor of statistical parameters features obtained by WT for faulty signals have been used as input for the ANN. If the wavelet coefficients are used as input to the ANN, it will result in rather large number of inputs posing difficulty for training and testing of ANN in connection with accuracy and speed. Therefore, the energy and dispersion factor of wavelet coefficients have been used as inputs to the ANN instead, in order to overcome this problem but retaining important feature of wavelet signals.
In some studies, Daubechies mother wavelet has good ability to capture the transient events and frequency feature extraction during fault in the power system. In this paper, the mother wavelet db3 with resolution level 3 used to obtain the coefficient of DWT for SLG-fault detection in unit generator-transformer. Some model for the original signal and parts of the coefficient with resolution level 3 of DWT db3 as illustrate in Figure 4 and Figure 5 respectively.
Fig.4. Original signals for SLG-fault current
Fig.5. Parts of DWT decomposition for current signals
(2). Features Extraction
The main idea of making a feature extraction is to reduce the amount of information, either from the original signals of from its transformation format. To reduce the number of ANN processing element, in this paper used a new approach of an energy and dispersion factor (R and STD value) of statistical parameters of wavelet coefficient for ANN input.
After getting the wavelet coefficient of the fault signals was obtained using signal decomposition, the next step in the extraction of the energy and dispersion factor of statistical parameters of the wavelet coefficients from WT in various fault simulations. Applying the energy and dispersion factor of each decomposition level, the numerical value of patterns can be obtained from analysed signal, as it is shown in Table 1 and Table 2 respectively.
Table 1. Energy feature vector of SLG-fault
.
Table 2. Dispersion factor feature vector of SLG-fault
.
The feature vector characteristics of each factor are then used as inputs for the ANN. In this case, 3000 signals were used for energy feature and 12000 signals were used for dispersion factor feature.
(3). ANN Trained and Verified
ANN has proven to very efficient in the field of classification. In this paper, the pattern recognition algorithms are used for classifying SLG-fault current and normal current condition in the unit generator-transformer. Pattern recognition is the process of training a neural network to assign the correct target classes to a set of input patterns. Once trained the network can be used to classify patterns it has not seen before.
MATLAB program has been developed for training process. The kinds of sample are divided into three namely training sample, validation sample, and testing sample. Training samples are presented to the network during training, and the network is adjusted according to its error. Validation samples are used to measure network generalization, and to halt training when generalization stops improving. Testing samples have no effect on training and so provide an independent measure of network performance during and after training.
The network has to detection of GF-fault at the various conditions of a unit generator-transformer. The inputs for network are extracting from dispersion factor of statistical parameters and energy of current details of wavelet coefficients. The WT is done to reduce the number of ANN processing element, and accordingly, it will reduce the time consumed for training and testing of the ANN. Moreover, it also helps to achieve high-performance detection.
The behaviour of the selected ANN depends on numerous parameters, such as the number of hidden layers, the number of hidden neurons, transfer function, initial weights and biases, training rule and training parameters. Table 3 shows the features of the constructed network. Two types of network were used for analysis with a number of different inputs. Model based on various input parameters as described in Table 4. 3000 sets of sample (70 % sets for training, 15% sets for validation and 15% set for testing) are used for energy wavelet coefficient network and 12000 sets of sample (70 % sets for training, 15% sets for validation and 15% set for testing) were used for R wavelet coefficient, STD wavelet coefficient or combined R and STD for ANN network.
Table 3. Features of the constructed network
.
Table 4. Models based on different input parameters
.
While training the network, energy and dispersion factor of wavelet coefficients pattern corresponding to varied conditions such as fault resistance, fault initiation time, and various generator grounding method are used. The targets for normal currents condition are trained to be ‘0‘, and the target for SLG-fault currents are trained to be ‘1’. Target vector is assigned value ‘1’ or ‘0’ according to the network condition. Threshold is set at 0.5. I.e values above 0.5 are treated as ‘1’ and values below 0.5 are treated as ‘0’. Once performance goals are met, an unknown pattern is applied to verify whether the network is trained properly or not.
The designed ANN is trained for various training patterns of normal and SLG-fault conditions. Various architectures were attempted to arrive at the final architecture with a goal maximum accuracy. After much experimentation, for ground fault detection six different architectures were developed and used for training. After enough experimentation, it was inferred that the architecture with one hidden layer of 20 neurons and one output was giving the optimum results. The goal of 0.0712 error is a achieved in 184 iterations during 27 seconds. Figure 6 Shows the graph between training performance and number of iterations to train the designed 4-20-1 of ANN structure.
Fig.6. Best validation performance energy as input for 4-20-1 of ANN structure
The performance of trained network for various architecture can be measured, to some extent, by the errors on the training, validation, and test sets. Comparison of mean squared error (mse) parameters of a pattern recognition model in various ANN structures as illustrate in Table 5.
Table 5. Mean squared error parameters of pattern recognition models for various network structures
.
From the Table 5, it appears that the case-1 result valid performance for testing SLG-fault signals than the case-2, case-3 and case-4. By using energy as ANN input, detection of the SLG-faults on the generator unit transformers was accurate enough in differentiating the SLG-fault and un-fault for a unit generator-transformer compared to R and STD value as input ANN.
Conclusion
This paper has presented a novel approach for SLG-fault detection at the unit generator-transformer. Regarding the ANN performance, the errors in the SLG-fault detection of ANN were under 1 %. In this paper, analysis of energy wavelet coefficients successfully applied to distinguish SLG-fault at the unit generator-transformer. The statistical parameters involved calculating the dispersion factors (R and STD value) of DWT were available to detect the GF.
REFERENCES
[1] Omar A.S, Youssef. Online Application of Wavelet Transforms to Power System Relaying. IEEE Transactions on Power Delivery 2003; 18: 1158-1165. [2] IEEE Std C37.102™-2006, IEEE Guide for AC Generator Protection. [3] J.C.Das. Power System Relaying. Wiley Encyclopedia of Electrical and Electronic Engineering, 1999. [4] A.R.Sultan & M.W.Mustafa. Ground Fault Protection Methods of a Generator Stator. Przeglad Elektrotechniczny 2013; 10: 225-229. [5] Silva.K.M, Souza.B.A, Brito.N.S.D. Fault Detection and Classification in Transmission Lines Based on Wavelet Transform and ANN. IEEE Transactions on Power Delivery 2006; 21 : 2058-2063. [6] Amir T, Mohammad-Reza M., & Abdolreza R. Fault Location Techniques in Power System based on Traveling Wave using Wavelet Analysis and GPS Timing. Przeglad Elektrotechniczny 2012; 6 : pp.347-350 [7] H.Zhengyou, G.Shibin, C.Xiaoqin, Z.Jun, B.Zhiqian & Q.Qingquan. Study of a new method for power system transients classification based on wavelet entropy an neural network. Electrical Power and Energy Systems 2011; 33: 402-410. [8] Safty S.E, El-Zonkoly A. Applying wavelet entropy principle in fault classification. Electrical Power and Energy System 2009; 31: 604-607. [9] Pittner.S, Kamarthi.S.V. Feature Extraction From Wavelet Coefficients for Pattern Recognition Tasks. IEEE Transactions on Pattern Analysis and Machine Intelligence 1999; 21: 83-88. [10] Rao.P.V.R, Gafoor SA. Wavelet ANN based stator ground fault protection scheme for turbo generators. Electric Power Components and Systems 2007; 35: 575-59. [11]Rahman, M.A, Ozgonenel O & Khan M.A. Wavelet transform based protection of stator faults in synchronous generators. Electric Power Components and Systems 2007; 36: 625-637. [12]Baqui I, Zamora I, mazon J & Buigues G. High impedance fault detection methodology using wavelet transform and neural network. Electrical Power System Research 2011; 81: 1325-1333. [13]S.Changqin, Wang.D, Kong.F & Tse.P.W. Fault diagnosis of rotating machine based on statistical parameters of wavelet packet paving and a generic support vector regressive classifier. Measurement 2013; 46: 1551-1564. [14]Chul-Hwan Kim, Hyun Kim, Young-Hun Ko, Sung-Hyun Byun, Raj K. Aggarwal and Allan T. Johns. A Novel Fault-Detection Technique of High-Impedance Arcing Faults in Transmission Lines Using the Wavelet Transform. IEEE transactions on power delivery 2002; 17. [15] Robi Polikar. The Story if Wavelets. Iowa State University [16]Morchen, F. Time series feature extraction for data mining using DWT and DFT, Technical Report, No.33, Department of Mathematics and Computer Science, University of Marburg, Germany, 2003 [17]Pham,T.V & Kubin,G. DWT-based classification of acousticphonetics classes and phonetic units. In proceeding of ICSLP’04 South Korea, 2004: 985-988. [18]Ekici, S., Yildirim S., Poyraz M. Energy and entropy-based feature extraction for locating fault on transmission lines by using neural network and wavelet packet decomposition. Expert Systems with Applications 2008; 34: 2973-2944. [19] Viljoen C. Elementary Statistic Vol. 2 Pearson South Africa, 2000 [20]Baqui I, Zamora I., Mazon J., & Buigues G. High impedance fault detection methodology using wavelet transform and artificial neural network. Electric Power System Research 2011; 81: 1325-1333. [21] MATLAB reference manual. The Mathworks Inc, 2012 [22]Othman,M., Mahfout,M., & Linkens,D. Transmission line fault detection, classification, and location using an intelligent power system stabiliser. IEEE Int. Conf. Elect. Utility Deregulat 2004; 1: 360-365. [23]Coury,D.V., Oleskovicz,M., & Aggarwal,R. K. An ANN routine for fault detection, classification, and locating in transmission lines. Electric Power Component System 2002; 30: 1137-1149. [24]Reaz, M., Choong, F., Sulaiman, M., Mohd-Yasin, F., & Kamada, M. Expert system for power quality disturbance classifier. IEEE Trans. Power Delivery 2007; 22: 1979-1988. [25]Al-Shaher., M. Saleh, A.S & Sabry, M.M. Estimation of fault locating and fault resistance for single line-to-ground faults in multi ring distribution network using artificial neural network. Electric Power Component System 2009; 37: 697-713. [26] Jiabin Z., Ju T., Xiaoxing Z., & Jiagui T., Pattern recognition for partial discharge in GIS based on pulse coupled neural networks and wavelet packet decomposition, Przeglad Elektrotechniczny 2012; 5b : pp.44-47
Authors: Ahmad Rizal Sultan, Politeknik Negeri Ujung Pandang, South Sulawesi, Indonesia 90245, E-mail: rizal.sultan@poliupg.ac.id Mohd Wazir Mustafa, Faculty of Electrical Engineering, Universiti Teknologi Malaysia(UTM), Skudai, Malaysia 81300. Makmur Saini, Politeknik Negeri Ujung Pandang, South Sulawesi, Indonesia. Ahmad Gaffar, Politeknik Negeri Ujung Pandang, South Sulawesi, Indonesia.
Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 94 NR 12/2018. doi:10.15199/48.2018.12.07
Published by Anna ZIELIŃSKA, AGH University of Science and Technology
Abstract: The article presents the infrastructure for charging electric vehicles, their development and the types and methods of charging. The paper presents the results of testing the level of vehicle charge depending on the route traveled and its dynamics. The test results show the change in the level of discharging the battery from the route length.
Streszczenie: Artykuł przedstawia infrastrukturę ładowania pojazdów elektrycznych, ich rozwój oraz rodzaje i sposoby ładowania. W pracy przedstawiono wyniki badań poziomu naładowania pojazdu w zależności od przebytej trasy i jej dynamiki. Wynik badań pokazują zmianę poziom rozładowywania baterii od długości trasy. (Infrastruktura ładowania i badanie spadku poziomu energii akumulatora samochodu elektrycznego).
Keywords: electric vehicle, charging, charging infrastructure. Słowa kluczowe: samochód elektryczny, ładowanie, infrastruktura ładowania.
Introduction
The motoring and electricity markets have been separate sectors of the economy not so long ago. They operated independently of each other, they did not have a common recipient –– a common denominator. It resulted from the fact that they presented types of energy carriers – although in both cases they were fossil. Currently, the dynamically developing automotive and transport sector opens up a whole new dimension of usability. The transport sector is responsible for about 30% of the total final energy consumption and for about 25% of harmful gas emissions [1]. One of the ways to reduce this share is to replace traditional vehicles with internal combustion engines (ICE) with battery electric vehicles (EV) and hybrid vehicles with an extended range with a battery (plug–in hybrid electric vehicle PHEV) [2].
Electric vehicles are much more energy–efficient and clean, i.e. they do not emit these impurities. Among other things, for this reason, as current sources state until 2030, half of the cars in the world will be electrified. 20% of cars sold in Europe until 2023. will have an electric motor. The total withdrawal of combustion cars from sale until 2030 is declared by countries such as Norway, the Netherlands and Germany [3].
To achieve these results and maintain the current upward trend, electric vehicles must be widely used in the future (Fig.2.). Although they still have a small market share, there is an increasing interest in this type of technology. This is achieved by overcoming their traditional bottleneck, which provides short range, high price (these two are mainly related to the battery) and lack of charging infrastructure, also of quick type [4].
To illustrate the main “deceleration of expansion” of the electric vehicle in the simple comparative analysis of EV and a diesel–powered car, the cost of purchase, annual operating costs, replacement cost of batteries (after 8 years for EV), fuel cost, electricity costs, and the assumed monthly distance covered on the level of 1500 km [5].
As can be seen from the graph (Fig. 1.), electric vehicle is more expensive to use today than traditional drive. However, the important fact is that EV has a much lower increase in costs over time –– it is more stable, of course, after an agreed period, there is a drastic increase in costs caused by the replacement of batteries, but nevertheless, in the longer term, its use looks appealing.
Fig.1. Running costs of an electric vehicle and a car with an internal combustion drive [5]
Fig.2. Car fleet in Poland [6]
The second factor limiting the interest in the electric vehicles is the lack of charging infrastructure. Currently, for the vast majority of EV holders, the charging process takes place at home, by connecting to a household power grid. The average charging time is so long that charging takes place at night and the small range of the battery limits long journeys.
Guided by the above–mentioned factors, the work presents charging methods for electric vehicles, infrastructure, future charging possibilities and the current application of the solution. The battery capacity and discharging tests are shown. One of the most popular cars in the EV sector – the Fiat 500e – was used for the analysis. Investigating the driving dynamics and its length shows the dependence of battery discharge and charging times
The infrastructure of charging electric vehicles
The appearance of electric vehicles permanently in the public space will mean a drastic change in consumer behaviour by offering them a new quality of movement – quiet, dynamic and ecological. However, to make this happen, the previously mentioned development of charging infrastructure for EV is necessary [7].
There is currently a division into three groups of charging stations for EV. The power level at the charging point has a significant effect on the battery charging time of the EV. We currently distinguish:
• I –the charger fits inside the car. From the distributor, alternating current is sent from a standard 230V single–phase socket. The converter power to be obtained is limited to 2 KW, which results in charging the battery depending on the capacity from 11 to 14 hours.
• II – The charger fits inside the car. The vehicle is loaded with alternating current, one or three–phase. The power can reach up to 20 KW, which means that the charging time is reduced to 2 – 3 hours,
• III – In this case, the charger is outside EV. The vehicle’s battery terminals are connected to a special connector located on the vehicle. It requires a DC power supply. The system’s power reaches up to 50 KW. This method allows you to charge up to 80% of the battery capacity in just 15 – 30 minutes, and the battery is fully charged in 1 hour [8].
Fig.3. Annual increase of charging points (forecasts for Poland) [6]
With the development of the market there will be various types of chargers for electric cars. One of them will be ultra–fast charging stations for DC electric cars with 100 kW and 300 kW. Their purpose will vary depending on where they are placed (Fig.3.). On highways, expressways, usually at hotels, restaurants or service areas, stations with high powers will be installed, for direct current, where the most important factor will be time, not the price of the service. In this case, customers will be willing to stop for 30–45 minutes to replace the battery for another 200––300 km [7]. It should be emphasized that even if the stations belong to many charging operators – CPO (Charging Point Operators), they will be associated with consistent settlement systems, enabling each client to use them, e.g. by means of a mobile application or RFID card, to enable moving over long distances in the country and in Europe. The situation was different in cities where cars travel a lot smaller distances, more often they park and can be loaded during overnight stays. The city infrastructure is developed primarily at freestanding stations (available in public or private areas, e.g. in garages of underground housing estates), equipped with slow or medium–speed chargers – between 11 and 22 kW AC.
A completely different concept, in contrast to the charging of contact electric vehicles, is charging wirelessly. For many, free of defects and providing unlimited range when electrified roads. In this type of charging, the most promising solution is the use of energy transfer on the principle of magnetic induction. A system composed of two coils, one in the vehicle of the other at the stopping place, magnetically coupled and forming a transformer with a large air space. The coil located in the transmitter generates a variable electromagnetic field, while in the coil placed in the vehicle, under the influence of this field, a variable electromotive force SEM is created. The energy after conversion in the charger charges the batteries [8]. Such a system is very simple to use and at the same time resistant to external factors [2]. Most often wireless power systems are used to power machines on production belts, however, for electric vehicles, such a system was also developed at the end of the nineties by General Motors.
There is also another way of wireless charging, based on the principle of electromagnetic resonance. The resonance system is mounted in the vehicle. EV charging occurs after the electromagnetic resonance of the transmitter and receiver is synchronized. This method is so much better than during charging there is no need for precise positioning of the system components. A big plus is a more efficient energy transfer. It is possible to transmit power of 3.3 kW at a distance of 20 cm, with losses of only 10%. These systems are lighter and much smaller than induction systems. The magnetic resonance charging method is cheaper, easier to build and safer compared to other wireless methods [8].
The classification and evaluation of wireless charging systems have:
• the power of the system that determines the duration of the loading process • the acceptable distance between the surface of the ground and the location of the system in the vehicle, • energy conversion efficiency, determined between the power supply network and the battery terminals • tolerance in positioning the vehicle on the parking spot, • vehicle dimensions and weight.
Another way to charge an EV that is considered but used only in a pilot manner is charging by changing the battery. The process involves replacing a discharged battery with a charged one, and charging takes place outside the vehicle. Battery replacement is to take place in a specially constructed station, which will be fully automated, and the entire process will be supervised and performed by robots [8]. In the future, the entire battery replacement process will take no more than one minute. However, as of today, the price of such a service has not been provided.
A separate aspect connecting directly with the infrastructure for charging electric vehicles will be the settlement of the charging process. The charging price will probably vary depending on the charging time, location and type of charger. This is also a big challenge for legal regulations on this topic.
In the future, science will probably create other ways of loading. Although today we would like the “fuel” that is electric energy to be available in such a way as to ensure driving safety in the context of the distance covered. This change in the way of refuelling will be both a challenge for customers and an opportunity for owners of residential and commercial properties to meet this emerging need. Ultimately, charging stations for EV will become common, democratization will take place in access to them, especially if the business or individual clients invest in their own renewable energy sources [7]
Battery discharge measurements
In addition to charging infrastructure, another issue in the context of the use of electric vehicles is their operation, i.e. energy consumption [9]. In the European Union in the approval tests, the energy consumption of electric vehicles is determined in accordance with the procedure described in UNECE Regulation No. 101. The vehicle is tested on a chassis dynamometer in the NEDC test, this test simulates urban and extra–urban driving [10]. In order to increase the level of information on the properties of electric vehicles, other tests are also performed in running tests, corresponding to different traffic conditions, as well as in traction conditions, during the actual use of the vehicle [11] [12].
To describe the energy and economic properties of electric vehicles, the concepts characterizing energy efficiency and consumption are used.
For electric vehicles without braking energy recovery, the efficiency system is defined as follows:
• drive efficiency
.
• battery charging efficiency
.
• general efficiency
.
where: NT – the power of the electric drive of the car, NR – power resistance, NCH –battery charging power.
For an electric vehicle with braking energy recovery, the efficiency system is defined as follows:
• drive efficiency
.
• efficiency of braking energy recovery
.
where: NB– braking power of the electric machine, NU – braking energy recovery power.
Road energy consumption is described as a derivative of the energy consumed relative to the distance travelled by the vehicle. For an EV without braking energy recovery, the road energy consumption is:
.
where: s – expensive vehicle, L(T)(s) – work of electric vehicle drive as a function of the road.
For an electric vehicle with braking energy recovery, the road energy consumption is:
.
where: LU(s) – regenerative braking energy as a function of the road [12].
The paper presents the results of an examination of an EV Fiat 500e model in urban conditions. The aim of the study was to assess the road use of energy, the battery power of its linear or non–linear decrease in terms of the number of kilometres travelled and the analysis of battery consumption in different driving dynamics and in the use of other systems existing in the car (such as air conditioning). Below is the technical data table of the car.
Table 1. Data for an electric car Fiat model 500e
.
The Fiat 500e is equipped with a traction drive with a maximum power of 83 kW, making it one of the more dynamic electric vehicles available on the market. Disputes for such a small car is also placed under the seats of a pack of batteries, accumulating 24 kWh of energy. To charge the batteries, Fiat decided to use only a 6,6 kW onboard charger. The car was produced in 2015 in Mexico and imported to Poland from California. At the beginning of the tests, the meter’s mileage was 56335 km. The car roamed the routes in the city in the summer season at ambient temperatures from 22 to 27 oC.
During the tests, the battery level before and after the route and the length of the route itself was checked. The analysis of the data collected during the tests shows that the considerable mileage of the car and hundreds of charging cycles did not negatively affect the battery condition in the car. The battery still retains its initial capacity. The tests were conducted in conditions of normal car use, in standard quiet and dynamic conditions of driving, in traffic jams, with the use of air conditioning and without. The main goal of the research was to see if the battery maintains a linear power drop or in higher battery power ranges the range of the car is greater. For the tests, a car with a considerable mileage was intentionally selected to eliminate the “new battery” syndrome. After analyzing the data, there was no significant deviation in relation to the linear decrease in the range of the car along with the decreasing level of battery charge. The graph below (Fig.4.) presents all 17 measurements showing the number of kilometres driven in relation to the decrease in battery power expressed in percentage points.
Fig.4. Dependency of the route travelled and the level of battery charge drop in% points
For most of the measurements made, the ratio of the battery power drop and kilometres travelled ranged from 0,8 to 1,2. For several measurements, the results did not differ much from this level, the points marked with a square in the graph show a higher value of the power drop per one kilometre travelled. Such measurements belong to exceptionally dynamic routes and for the air conditioning system in operation. With such parameters of driving, faster battery discharge is observed. The point on the graph marked with a triangle shows a short ride in the traffic jam with the air conditioning set to the lowest temperature – for this route, there were about 2,6 points of the power drop per one kilometre travelled.
After the analysis, it was noticed that the length of the route is less important than its dynamics and driving style. For constant speed the power loss will be in a linear and predictable way while driving with a higher load on one battery charge, we will travel a longer route. Nevertheless, the energy saved is not significant enough to be a determinant and determines the driver’s behaviour.
Summary
Most electric vehicle is sold in Norway, France, Germany and the United Kingdom (these four countries account for 72.4% of all new registrations of EV in Europe [5]), almost in all Europe, the number of registered electric cars is growing every year. As the markets show, the development and popularity of electric vehicles is surprisingly growing year by year, but nevertheless most importantly can be recognized as the fact that:
• the current development priority for electric vehicles is the development of charging infrastructure,
• the decision on the purchase of an EV in addition to the price is decided by the range and the possibility of charging, in most cases only the newly created charging points of electric vehicles are able to influence the decision of the buyers regarding the type of drive,
• ensuring the possibility of loading, we eliminate the fears of limited distance that can be overcome,
• choosing a car with electric drive on a large scale will affect the ecology and the environment – the elimination of harmful gases and dust will absolutely improve the comfort especially in large urban agglomerations where the traffic volume is the greatest.
• Studies of energy consumption by electric vehicles are usually carried out under homologation conditions. In the case of tests carried out, it was tried to reproduce the conditions of normal car use as faithfully as possible, i.e. in the presented work it was shown that:
• to reproduce the everyday conditions of car use, the car was introduced into a traffic jam, subjected to smooth and smooth driving and dynamic driving with the use of on–board systems,
• research results showed a fairly low sensitivity of energy consumption to the vehicle traffic model,
• noticeable increased energy consumption from the battery was noticed while driving in a traffic jam, with the air conditioning system turned on and for dynamic and fast
• nevertheless, in all cases the average energy consumption per one kilometre of the route is satisfactory and gives the opportunity and perspective for the spread and popularization of electric vehicles on the roads.
LITERATURA
[1]. U.S. Energy Information Agency. International Energy Outlook 2014; (2014) [2]. Guziński J., Adamowicz M., Kamiński J., Infrastruktura ładowania pojazdów elektrycznych, Automatyka-Elektryka-Zakłócenia, 1/2014, (2014) [3]. Zielińska A., Skowron M., Bień A., Infrastruktura fotowoltaiczna do ładowania pojazdów elektrycznych — Photovoltaic infrastructure for charging electric vehicles, XXVIII sympozjum środowiskowe PTZE, ISBN10: 83-88131-99-0., (2018), 370–372 [4]. Nunesn P., Figueiredo R., Brito M. C., The use of parking lots to solar-charge electric vehicles, Renewable and Sustainable Energy Reviews, 66/2016, (2016), 679–693 [5]. http://samochodyelektryczne.org/wyniki_sprzedazy_aut_elektrycznych_w_europie_w_2017r_kraje_i_modele.html for 23.07.2018 [6]. Korolec M., Boleska K., Napędzamy Polską Przyszłość, 19.02.2018. [7]. https://www.muratorplus.pl/technika/instalacjeelektryczne/stacje-ladowania-samochodow-elektrycznychrodzaje-stacji-ladowania-sposoby-rozliczen-aa-AsuM-ME8B-93W8.html for 01.10.2018 [8]. Zajkowski K., Seroka K., Przegląd możliwych sposobów ładowania akumulatorów w pojazdach z napędem elektrycznym, Autobusy, 7-8/2017, (2017) [9]. Chłopek Z., Lasocki L., Badania zużycia energii przez samochód elektryczny w warunkach ruchu w mieście, Zeszyty Naukowe Instytutu Pojazdów 1(97)/2014, (2014) [10]. Raslavičius L., Starevičius M., Keršys A., K. Pilkauskas K., Vilkauskas A., Performance of an all–electric vehicle under UN ECE R101 test conditions: A feasibility study for the city of Kaunas, Lithuania. Energy, 55(15), (2013), 436–448 [11]. Lorf C., Martínez–Botas R.F., Howey D. A., Lytton L., Cussons B., Comparative analysis of the energy consumption and CO2 emissions of 40 electric, plug–in hybrid electric, hybrid electric and internal combustion engine vehicles. Transportation Research Part D, z. 23, (2013), 12–19 [12]. Wantuch A., Kurgan E., Gas P.: Numerical Analysis on Cathodic Protection of Underground Structures, in 2016 13th Selected Issues of Electrical Engineering and Electronics (WZEE), IEEE Xplore, (2016) [13]. Chłopek Z., Badanie zużycia energii przez samochód elektryczny,https://depot.ceon.pl/bitstream/handle/123456789/562/POL_2012_3_Badanie_zuzycia_energii_przez_samochod_elektryczny.pdf?sequence=1&isAllowed=y for 01.10.2018
Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 95 NR 1/2019. doi:10.15199/48.2019.01.38
Published by Mariusz ŚWIDERSKI, Michał GWÓŹDŹ, Poznan University of Technolog
Abstract. In the work the other approach to a photovoltaic system is presented. With regard to an improvement of reliability of a solar system, maintaining and improving system’s efficiency, authors proposed conception of the distributed PV system. This system consists of a set of individual small PV panels, while a single panel is connected to a low power converter, equipped with a pulse transformer. The converters work at a common energy container (battery). Thus, in case of failure of a single converter (or panel) the system is able to work properly with small only decrease of an output power, apart, maintenance of system operation is facilitated. In addition, due to all converters operate independently, individual environmental conditions (e.g. panel’s temperature) can be respected so system’s efficiency can be improved. In the work basics of system operation and selected system’s simulation model findings are presented.
Streszczenie. W pracy przedstawiono inne podejście do struktury systemu fotowoltaicznego. W odniesieniu do poprawy niezawodności działania systemu, obniżenia jego kosztów utrzymania i poprawy wydajności pracy, autorzy zaproponowali koncepcję tzw. rozproszonego systemu ogniw PV. System ten składa się z zestawu niewielkich paneli PV gdzie, pojedynczy panel jest dołączone do przekształtnika energoelektronicznego małej mocy, wyposażonego w transformator impulsowy. Transformator zapewnia mu indywidualną izolację galwaniczną. Konwertery pracują na wspólny zasobnik energii w postaci baterii LiION. Tak więc, w przypadku awarii pojedynczego konwertera (lub panelu), system może działać dalej, przy zmniejszonej mocy wyjściowej. Ułatwiona jest również obsługa (naprawa) systemu. Ponadto, ponieważ konwertery działają niezależnie od siebie, mogą brane być pod uwagę indywidualne warunki środowiskowe dla każdego panelu – na przykład jego temperatura, czy stopień zaciemnienia. W niniejszej pracy przedstawiono podstawy działania systemu i wybrane wyniki badań modelu symulacyjnego pojedynczego panelu. Elektrownia solarna z rozproszonym systemem ogniw fotowoltaicznych
Keywords: big data system, flyback converter, photovoltaic cell, solar system. Słowa kluczowe: big data system, cela fotowoltaiczna, przetwornica flyback, system solarny.
Introduction
Currently, photovoltaic systems solutions consist of a single PV panel and a single converter or group of PV panels and a single converter. Taking into account mainly the improvement of the reliability of the solar system while, maintaining or even increasing the efficiency of energy conversion, a solution using a distributed system, i.e. separation of the PV on practically single cells (or a few) and coupling it with low power electronics converters is proposed in [1]. This approach requires very sophisticated controlling and refreshing the already known structures of inverters, but also the control algorithms. In addition, there will be considered option to implement in converters modern transistors based on gallium nitride (GaN) material. In this case, it is required to develop a new method of control of power devices in the structure of power converters [2]. The characteristics of GaN transistors let achieve a much higher switching frequency – in relation to Si or even SiC devices. However, at the present stage of the work low loss, ultra fast power MOSFETs are taking into account.
The main goal of the research is to increase the reliability of the system and the resultant efficiency of the entire system, so that it could be achieved the greatest efficiency of converting solar energy into electricity. For this purpose mini converters with a total power equal to the power of a single photovoltaic panel to which, they are connected. Such a solution can increase the reliability of the system, because in the event of failure of one of the DC/DC converters or PV panels, the device can continue to operate with less power, and service of devices can take place at a convenient time, moreover converters operate sequentially with regard to their working conditions e.g. temperature, in order to achieve the operational wear to be uniform. In the studies will be verified the validity of the thesis that converters in distributed system may have a higher conversion efficiency of electricity.
The proposed solar energy conversion system includes an energy container based on LiION cells. This one is common for an entire system and preserves a continuity of energy supply for a consumer e.g. during time of reduced solar radiation. This work presents the initial stage of the project of the solar power system.
Basics of system
The general conception of energy generation system based on distributed PV panels is shown in Fig. 1.
Fig.1. Block diagram of energy generation system based on distributed PV panels
The system consists of the following blocs: set of low voltage photovoltaic panels (PVP) in the number of N , set of low power converters (CNV), where an individual converter is coupled with a single PVP, energy container (EC) – based on LiION cells, power grid side converter (GCN), coupling EC with a power grid, and global control block (GCB), including dedicated big data control algorithm (BDCA). The individual converters are connected to GCB via isolated data-control bus.
The CNV block includes a flyback converter with local controller (LCT). The LCT monitors: PVP’s output voltage ( uPV ), PVP’s output current ( iPV ), and temperature (TPV ) of the panel. On base of these quantities the LCT realizes the Maximum Power Point Tracking (MPPT) algorithm.
In the extended version of the system (Fig. 2) the CNV block contains a larger number (i.e. K) of lower power flyback converters, being connected in parallel. If power transferred by a single flyback converter reaches its nominal capacity, another converter is switched on, etc.. Thank to this they operate in conditions, being close to nominal ones. As a consequence, it is expected, that overall system’s efficiency will grow. Moreover, an order of turning on of another converter is not fixed. This one bases on pseudorandom algorithm. So, the expected system’s lifetime should increase.
Fig.2. Diagram of CNV block in its extended version
The essence of system’s control is to provide a matched power to a load, while maintaining reliability at a high level for the entire structure. The algorithm is characterized by high dynamics of operation, while maintains all principles involved in design of converters for solar systems, e.g. MPPT [3]. Solar radiation falling on photovoltaic panels is disturbed by many factors, e.g. polluted air and, sediments on the panels or even cloud cover. Therefore, the density of luminous flux for particular photovoltaic panel is not uniform. Due to this aspect of the application of the control using MPPT algorithm in a distributed system causes the individual and more effective adjustment to the maximum received power from each photovoltaic panel. In addition, the algorithm receives information about both operation of the converters and state of the loads.
The data used in the analysis is based on Big Data algorithms, to effectively predict the behavior of both the generation side and the load side. The justification for the selection mechanisms of Big Data [4, 5] is dictated by the presence of a large diversity in the collected information, and also to achieve adequately fast prediction of events (i.e. an increasing demand for power by the load or set of loads, the decline in power generation by reducing the light radiation generated by the sun or the temperature increase of photovoltaic cell) in the operation of the entire system, there is a need to collect vast amounts of information – this would include eclectic parameters measured in virtually every component of the system and the load terminals, and ending with the data collected from the environment such as temperature and solar irradiance.
Big Data systems according to the 4V [6] model should meet the following requirements:
• volume – have large amounts of data, • velocity – characterized by high variability of data, • variety – consists of a large variety of data, • value – system should collect data of significant values.
In order to meet these requirements, a set of algorithms responsible for the operation of the system was developed. The first of the described algorithms is the algorithm controlling the system startup section – shown in Fig. 3. The system should operate independently. Therefore, the detection of the number of DC / DC converters with which it works is crucial in the start phase. If this process is completed successfully, the system will proceed to database analysis. However, if the system does not detect the connected converters or does not communicate with them, the user will be informed about the system error.
During the boot sequence data analysis is mainly based on counting the number of records in the database. On this basis, the average polling time of system components is determined. In case of slow-changing systems, this limits the size of the database by the reduced frequency of system polling. In other case for processes with significant dynamics, the amount of data will increase accordingly in order to be able to correctly calculate system behavior patterns at a later stage. If the size of the database is not greater than the accepted minimum ( Rmin ) then the system accepts a random (from the range) response frequency, which the system will be able to correct.
Fig.3. System’s startup sequence algorithm
The second described algorithm shown in Fig. 4 is the algorithm responsible for data acquisition.
Fig.4. Data acquisition algorithm
The main task of the algorithm is to collect data and save it to the database. Then, based on the data collected in the database, the average polling time of system components is updated. Next, the demand for system power is determined. If this demand is lower than the power of the whole system then the algorithm can turn off the least effective elements of the system. The reason for low efficiency may be shading, dirt or breakdowns. This algorithm operates in a closed loop and its operating frequency is automatically selected.
The last presented algorithm is the algorithm responsible for deleting records from the database [5]. This one is shown in Fig. 5.
Fig.5. Database cleaning algorithm
In parallel with the control procedure, the system ensures that the database contains only relevant records. Hence, cyclically selected parameters records are read from the database and compared to averaged parameter values. If the average value differs from the analyzed value by the pre-determined degree of accuracy ( e ) – then the record is deleted. If the analyzed value is unique, i.e. it differs significantly from the averaged value, the analyzed record remains in the database.
The next step in the research of the distributed system of converters will be to check the different methods of arranging photovoltaic panel modules, e.g. in the shape of a paraboloid, a sphere section or other hypersurfaces. In the case of a flat arrangement of modules, the energy of reflected radiation is lost. However, in the proposed solution it is possible to reuse part of this energy. Arranging the modules in the shape of a hypersurface additionally contributes to averaging the value of energy received during the day. In comparison to the classic system, the maximum energy consumption is clearly higher than the average and falls at noon. Laying the panels in the shape of a hypersurface additionally eliminates the need for an expensive mechanical system to keep up with the sun (“sun-follower”) [3, 7] or expensive optical elements.
Simulation model studies
At the present project’s development stage simulation studies devoted to the converter in the CNV block (in ORCAD/SPICE environment) were conducted.
The simulation model of this block (in simplified form) is shown in Fig. 6. Basic parameters of the model are as follows:
• PV panel output voltage: 2.4÷2.7 V, • PV output current (max): 2.0 A, • energy container voltage (nominal value): 48 V, • switching frequency of MOSFET in CNV block (max): 100 kHz.
Functionality of main blocks in the simulation model is as follows: PVP is PV panel’s model (PV contains three small PV cells connected in series), SW is the power switch model, and CTB is the control block model. The pulse transformer’s model (TX1) was based on a real planar transformer with ferrite core type EEQ30 (N97 material) from TDK. The power switch model was based on modern OptiMOS™ 5 100 V power MOSFET type BSZ146N10LS5 manufactured by INFINEON [8]. Thanks to suitable design of the pulse transformer (both a flat core and multilayer windings) its coupling factor is close to 1 ( k ≅ 0.995). Thus, taking into account, that magnetizing inductance is equal to ~25 µH, the leakage inductance (associated with the primary winding) is equal approximately to 100 nH. As a result, a form of the snubber circuit (R56-C31) can be very simple.
Fig. 6. Block diagram of CNV simulation model
Fig.7. Waveforms of voltages and currents in CNV simulation model, while converter operates in: a) DCM mode, b) CCM mode
In Fig. 7 waveforms of selected voltages and currents in the simulation model are shown. As can be observed, a magnitude of primary current in the DCM is equal to 0.75 A, whereas in the CCM is equal to 1.8 A. A value of power transferred to the energy container was in the range 0.8÷3.0 W – depending on the mode of operation of converter. As it is shown, magnitudes of voltage spikes, in the power switch transient states, are relatively low. In a result the value of energy, dissipated in the snubber circuit, is negligible from point of view of converter efficiency.
Conclusions
The entire study aims to determine the level of reliability of the proposed solution and check how significant is impact of the efficiency of power electronic converters on the overall efficiency of solar energy conversion system. At present stage of system developing the general structure of control algorithms was proposed. With regard to an improvement of reliability of a solar system, maintaining and improving system’s efficiency authors used the distributed PV system consists of a set of individual small PV panels, while a single panel is connected to a low power (micro) converter, equipped with a sophisticated pulse transformer. The converters work at a common energy container (e.g. LiION battery stack). Thus, it is expected, that in case of failure of a single converter (or cell in panel) the system will be able to work properly with small only decrease of an output power, apart, maintenance of system operation is facilitated. Also, findings may open a new way for use of the GaN transistors in systems, where the energy conversion efficiency is crucial, and en energy’s source is limited in power or difficult and expensive to operate.
REFERENCES
[1] Z. Jin, M. Hou, F. Dong, and Y. Li, A new control strategy of dc microgrid with photovoltaic generation and hybrid energy storage, Power and Energy Engineering Conference (APPEEC), (2016) IEEE PES Asia-Pacific, 2016. [2] Gwóźdź M., Matecki D., Power electronics controlled voltage source based on modified Sigma-Delta modulator, Proceedings of the 2016 IEEE International Power Electronics and Motion Control Conference (PEMC), Bulgaria, Varna, 25-30 September (2016), ISBN: 978-1-5090-1797-3, pp. 186-191. DOI: 10.1109/EPEPEMC.2016.7751995. [3] Degeratu S., Alboteanu L., Rizescu S., Coman D., Bizdoaca N., and Caramida C., Active solar panel tracking system actuated by shape memory alloy springs, Applied and Theoretical Electricity (ICATE), (2014) International Conference on, ISBN: 978-1-4799-4161-2. [4] Swartz R.A., Lynch J.P., Zerbst S., Sweetman B., and Rolfes R., Structural monitoring of wind turbines using wireless sensor networks, Smart Structures and Systems 6, 114, (2010). [5] Najafabadi M. M., Villanustre F., Khoshgoftaar T. M., Seliya N., Wald R., and Muharemagic E., Deep learning applications and challenges in big data analytics, Big Data, vol. 2, no. 1, March (2015). [6] Erhard R., The case for holistic data integration, East European conference on advances in databases and information systems, Berlin: Springer; (2016). [7] Zamojski W., Theory and technique of reliability, (in Polish), Wrocław University of Technology, Wrocław, (1976). [8] Infineon product page: https://www.infineon.com/cms/en/product/power/mosfet/. Accessed: December 2017.
Authors: mgr inż. Mariusz Świderski, E-mail: Mariusz.Swiderski@put.poznan.pl, dr hab. inż. Michał Gwóźdź, Email: Michal.Gwozdz@put.poznan.pl, Politechnika Poznańska, Instytut Elektrotechniki i Elektroniki Przemysłowej, ul. Piotrowo 3A, 60-965 Poznań.
Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 95 NR 2/2019. doi:10.15199/48.2019.02.11
Published by Rakesh Kumar, EE Power – Technical Articles: Trends in Electric Vehicle Fast Charging, February 02, 2023.
Ultra-fast charging methodology, ultra-fast charging station architectures, and improved battery technology are some promising trends for fast-charging electric vehicles.
EVs hold the potential for decarbonizing the transportation sector. But the crucial impediment to the adoption of electrified transportation is the charging time taken by an EV. A fuel-based vehicle takes only 15 minutes or less to refuel. Therefore, to encourage the use of EVs, the charging time is expected to match that of fuel-based vehicles.
Fast charging is key to alleviating range anxiety issues associated with EVs. Various trends are observed in the fast charging of EVs, such as ultra-fast charging, higher battery capacity, and architectures for ultra-fast charging stations.
Ultra-Fast Charging is the Future
Figure 1 shows that the energy demand for EVs is set to go up drastically in the near future. The energy demand is considered in three emerging economies that will boost the adoption of EVs. The illustration in Figure 2 throws light on the importance of fast charging in the coming years. Fast charging is on the roadmap of every emerging economy to boost EV usage. The level 2 and DC fast charging will witness a surge in its share of the total energy demand as time passes.
The DC fast chargers are rated at 50 kW, marching towards ultra-fast charging. Most drivers prefer to charge the EV battery within 15 minutes which is very challenging from a technological perspective. Some commercially available EV models that capture such fast charging technology are Mini Copper SE, BMW i3, Hyundai Kona, Tesla Model 3, and Tesla Model S. The battery capacity ranges from 25 to 95 kWh, with a range starting from 180 km to 515 km.
Better Battery Technology
As EVs continue to evolve, their battery capacity is expected to increase. The advancement in power electronics technology alone cannot achieve ultra-fast charging. The current battery technology limitations also restrict how fast we can charge an EV. Energy capacity is one of the battery parameters to look out for in long-range EVs. Lithium-ion batteries are the most suitable for EVs of the many existing battery technologies globally. The lithium-ion batteries have a higher power and energy density compared to its counterpart. These features help in removing the EV range anxiety problems of the masses.
The material composition of electrodes used in the lithium battery is a key factor in deciding the energy density of the battery technology. Lithium-ion batteries can be charged in 15 minutes with an energy density of 150 Wh/kg or more with the latest state-of-the-art materials. CATL company has utilized graphite and lithium nickel manganese cobalt oxide (NMC) as an anode and cathode, allowing the battery energy density of 215 Wh/kg. Kokam is another battery manufacturing company with the same anode and cathode composition, offering a battery energy density of 152 Wh/kg. Enevate has utilized Si and NMC as anode/cathode compositions for an energy density of 350 Wh/kg.
Thermal management is another serious issue with battery management. As the energy and power density of the battery increases, it is crucial to look out for proper thermal management practices. Overheating is a fundamental thermal management problem in an electric vehicle. The battery packaging is not designed properly to avoid cooling loss in the battery pack. Such an event leads to events where the battery catches fire, and the whole EV is engulfed in flame. EV battery performance is also affected by cold temperatures. Lithium-ion batteries tend to perform slowly in charging and discharging batteries when placed at very low temperatures. Hence, in such a case, a nominal battery temperature is necessary through mild heating without overheating the battery pack. A third issue prominent in the thermal management of EV batteries is thermal runaway conditions. It is a phenomenon where an increase in temperature aids in further heating up batteries if the temperature is not regulated properly.
Ultra-Fast Charging Station Architectures
The ultra-fast charging station needs to employ its unique architecture to enable fast charging of different EVs connected to it. These stations are most expected along long-distance highways where regular charging is necessary for EVs. Tesla’s fast charging stations consist of nearly 10 to 12 direct current fast chargers, each one bearing a capacity of 150 kW. Therefore, a typical fast charging station has to be rated at 1.5 to 1.8 MW capacity. It is advisable to draw the power from a medium voltage grid for such high power. The low-voltage grid cannot handle such high power, and it might also put an additional burden on the transformers used.
Figure 3 shows a conventional AC distribution network-based ultra-fast charging station. This architecture utilizes multiple AC to DC converters dedicated to each charging point. The AC distribution network is, at present, the most mature architecture, and it is easily viable commercially.
Figure 4 shows the DC distribution network-based ultra-fast charging station. This architecture is currently being researched for its efficiency and commercial viability. This architecture utilizes a simpler architecture with a reduced number of conversion stages. A single AC-to-DC converter is used, after which the DC power is distributed to all the charging points. A promising future ahead for the DC-based network configuration is the usage of Solid State Transformer in the initial conversion stage to replace the combination of medium voltage grid and AC to DC conversion stage. It helps improve the overall system’s efficiency with a battery control mechanism due to the use of power electronics technology.
Figure 5 offers greater insight into the different power electronics converter topologies of an AC distribution network-based ultra-fast charging station. It consists of AC-to-DC and DC-to-DC power conversions. The AC-to-DC conversion is also called the power factor correction stage, the first of the two stages of fast-charging power conversion. Three topologies commonly used for this stage are the Vienna rectifier, conventional 2-level voltage source rectifier, and multi-pulse rectifier. The common features of all these topologies are simplicity in design, higher reliability, and their ability to draw input currents with low harmonics distortion. The second stage is the DC-to-DC conversion stage, where three commonly used topologies are the Half-Bridge LLC, interleaved buck converter, and dual active bridge converter. These topologies can provide galvanic isolation between the EV and grid to enhance the reliability of the whole charging station.
A modular structure of power electronics converters in the DC fast charging is useful in many ways. It helps distribute the voltage and current stress equally among the different modules. Each module can cater to the unique voltage level demand. Therefore, modularity allows for different voltage and power handling capacities within an ultra-fast charging station. As the modules are spaced at a reasonable distance from one another, it allows for proper cooling of each module. In the future, the power handling capacity of an ultra-fast charging station can be increased or decreased by adding or deleting the individual modules.
Key Takeaways of Electric Vehicle Fast Charging
Electrified transportation is witnessing some trends in fast charging. The article has highlighted and briefly explained some important trends. Some of the takeaways of the article are as follows.
• Ultra-fast charging is the need of the hour to facilitate charging the EV battery in the least possible time possible.
• Battery technology will play a pivotal role where the main challenge is increasing the battery energy and power capacity.
• Increasing the battery capacity also needs to address critical thermal management issues. Overheating, cold climatic environments, and thermal runaway conditions are key points in thermal management.
• AC and DC distribution-based ultra-fast charging stations are the two architectures that have a high potential to cater to the demands of fast charging. AC-based architecture is mature enough and is the ideal starting point.
• However, DC-based architecture is gaining popularity due to the emergence of Solid State Transformers, which can better control and simpler conversion stages.
Author: Rakesh Kumar holds a Ph.D. in Electrical Engineering with a specialization in Power Electronics from Vellore Institute of Technology, India. He is a Senior Member of IEEE, Class of 2021, and a member of the IEEE Power Electronics Society (PELS). Rakesh is a committee member of the IEEE PELS Education Steering Committee. He is passionate about writing high-quality technical articles of high interest to readers of the EE Power Community. You can email him at rakesh.a@ieee.org.
Published by Dušan MEDVEĎ1, Zsolt ČONKA1 Marek PAVLIK1, Ján ZBOJOVSKY1,Michal KOLCUN1,Michal IVANČÁK1, Technical University of Košice, Department of Electric Power Engineering, Mäsiarska 74, 04001 Košice, Slovakia (1)
Abstract. This paper deals with the prediction of electricity generation in particular part of the network (island operation) where were considered various regimes of the wind power plant as a one of the power sources. The simulation network was created in Matlab/SimscapePowerSystem environment that consisted of rotating generators (for regulation of power due to fluctuated wind power generation) and wind power plant of variable energy generation and loads. There were considered the following wind power plant regimes: dynamic wind speed and dynamic load; dynamic wind speed and constant load; constant wind speed and dynamic load. From the all regimes were created prediction diagrams which form the day diagram of load.
Streszczenie. Artykuł dotyczy prognozowania produkcji energii elektrycznej w wydzielonej części sieci pracującej wyspowo, gdzie rozważano różne reżimy eksploatacji elektrowni wiatrowej jako jednego ze źródeł mocy. Sieć symulacyjna, opracowana w środowisku Matlab/ SimscapePowerSystem, składała się z wirujących generatorów (do regulacji mocy z uwagi na fluktuacje generacji wiatrowej) i elektrowni wiatrowej o zmiennej generacji energii i mocy. Rozważano następujące reżimy pracy elektrowni wiatrowej: dynamiczną prędkość wiatru i dynamiczne obciążenie; dynamiczna prędkość wiatru i stałe obciążenie; stała prędkość wiatru i dynamiczne obciążenie. Ze wszystkich reżimów powstały diagramy predykcyjne tworzące dobowy przebieg obciążenia. (Prognozowanie produkcji energii elektrycznej w pracy wyspowej dla różnych trybów generacji wiatrowej).
Keywords: wind power plant, off-grid network, flicker –effect. Słowa kluczowe: elektrownia wiatrowa, praca wyspowa, efekt migotania.
Introduction
This article presents the particular results of the simulation of the impact of various electricity sources on a small off-grid. Diesel generators and wind turbines have been used as power sources. From the point of view of electricity consumption, the effect of disconnection or connection of a large load on the system and the effect of a dynamically changing load is described. Multiple circuits have been simulated to verify some of the network phenomena. The main monitored variables included network frequency, voltage at the point of consumption, and power produced by sources.
To simulate these phenomena, the Simscape Power Systems, which is an extension of Matab Simulink, was used. Based on the simulation analysis, a simple solution was developed to reduce the impact of transient phenomena. Since simulated transient phenomena of a short nature, i.e. they take a short time, the designed simulations simulate the time interval within 1000 seconds, which is about 17 minutes. A short time interval has also been chosen because the results that are written in matrices have some accuracy and can be processed with current common computing techniques.
Description of the network model of components in the environment of Simscape Power Systems
In the Simscape Power Systems, several electrical machines are implemented. Many of these electrical machines can work in two states – as electricity generators or as motors, that is, as electric consumer appliances [1, 2]. Model of the synchronous machine with expressed poles was used. The synchronous generator is controlled by a hydraulic turbine combined with the PID control system and excited by the AC4A excitation system. The principal scheme of the G1 generator with a control and exciter system and generating output from the generator can be seen in Fig. 1.
Output of the synchronous generator is a three-phase voltage at the terminals of the machine A, B and C and the measurement output marked with the letter “m”. The measurement output includes a vector with measured signals: stator currents, stator voltages, rotor angle deviation, rotor speed, electromagnetic torque, output active power P, output reactive power Q, and so on. These signals receive feedback from the generator that is input to the exciter winding input and the hydraulic turbine with the control. The label data of the simulated generator are shown in Table 1.
Fig.1. Principal scheme of the G1 generator connection
Table 1. Data of the simulated generator.
.
In Fig. 2 is a model of a hydraulic turbine with PID control. This model has 5 inputs and 2 outputs. Inputs include reference speed, instantaneous mechanical speed, speed deviation, reference power and instantaneous power output. The output is the mechanical power Pm, which is also the input for the synchronous generator. In the mentioned model was set the reference speed ωef = 1 pu, and the inputs of the immediate mechanical velocity ωe and the velocity variation dω were connected. This regulation ensures the regulation of the synchronous generator at the nominal frequency fn = 50 Hz. Inputs of the reference mechanical power Pref and instantaneous power Pe0 are not connected. The circuit is set so that it does not take any feedback (or feedback from the gate output). It has been achieved that the turbine power was controlled only by rotor speed [7].
Fig.2. Hydraulic turbine diagram with PID control
Wind turbine
The wind turbine block with power-out transmission to the grid is considerably easier than a block of PV field. The wind turbine input is the wind speed reported in m·s–1 and a Trip connector. The wind speed for this model was retrieved from a text file. Trip connector serves to simulate the turbine protection system. Its input may be a logical zero or one. If at input port is logical zero, the wind turbine is in operation and when at input is logical one, the turbine is disconnected. The wind turbine can have several protections. First of all, it is a wind turbine disconnection when there is slow/fast wind, but also overcurrent protection, undervoltage protection, overvoltage protection, or protection, acting in the unbalanced current or voltage.
The wind turbine output is a measuring port that contains the voltage and current at turbine terminals A, B and C, turbine output P and Q, turbine rotor speed, mechanical torque, and so on. The wind turbine used in this article includes, in addition to the turbine, also an asynchronous engine that generates the electrical energy. In Fig 3 is shown the characteristic output of the designed turbine at different wind speeds.
Fig.3. Characteristic turbine power at different wind speeds
Since during the simulation the disconnection of the wind turbine caused a mathematical error, the block “Check static range” was added to the scheme. This block stops the simulation if a wind speed is read at a speed that is not in the work range, and Matlab shows error message. The wind turbine operating range is in ranges from 4.5 m·s–1 to 12.5 m·s–1. The basic wind speed for the model turbine was set to 9 m·s–1.
Definition of Loads
Simscape Power Systems offers several types of loads. In this article, three-phase serial RLC load and three-phase dynamic load were used. For both loads, the combined nominal voltage and nominal frequency of the network were entered.
Fig.4. Wind turbine block with power delivery to the system
For a static three-phase load, PQ power was entered, which may be the same or specific for each load in all three phases. A static three-phase load contains also a voltage and current measurement that is optional [3, 4, 6].
For dynamic three-phase load, the PQ power was entered at the beginning of the simulation. The PQ power of a dynamic load can be controlled by an internal control that controls the amount of dynamic load based on the positive sequence voltage component. If external control of power source is used, performance can be read from a file, and controlled by an external handling. The dynamic load contains also a measuring terminal „m“, the output of which is a vector with a positive-sequence voltage component, an active power P and a reactive power Q [5, 10].
Fig.5. Loads in Simscape Power Systems (3-phase series RLC and 3-phase dynamic load)
The loads were read using a Matlab script. In Fig 6, a proposed load block for the supply point A is shown. On the left, the load A1 and line A2 are shown, which are connected to the system via a three-phase circuit breaker and a power line simulated by the impedance Ra and La. Line A2 consists of a purely ohmical load, because the dynamic load line A1 cannot be connected in series with the inductive element of the three-phase line, which is the supply point connected to the system. On the right, the reading of block of line load A1 is displayed. If init_const = 1, the load, i.e. line A1 is set according to the vector from a text document. If init_const = 0, the load is set to the constant value, which is set in the text document for time t = 0. Current and voltage measurements were performed on bus-bar A. During the simulations, four consumption points A, B, C and D were considered. Each of these consumption points represents a part of the network. In some simulations, only static three-phase consumption points were used that were disconnected by a three-phase switch.
Measurement in Simscape Power Systems
In the simulations, electrical quantities were measured at selected locations in the network. Phase currents and voltages were measured using three-phase V-I measuring blocks, which were placed before loads and before the generators, resp. other sources. The measured output is the sinusoidal voltage/current depending on the time that has to be converted to the effective value (for comparison purposes). The scheme for measuring of the particular variables at the output of the G1 generator is shown in Fig. 7. The RMS current and voltage values for the L1 phase and the active and reactive power in the L1 phase were calculated from the measured currents and voltages.
Fig.6. Designed load block with control
Fig.7. Scheme for control and measuring of monitored quantities
Model of a steady-state off-grid network
Fig.8. The steady-state model diagram
In the off-grid steady-state model, the main aim was to point out that if no changes were made to the scheme and the correct initialization conditions were set, the network’s frequency did not change and was 50 Hz. The phase voltage in phase L1 is equal to the portion of the line-to-line voltage and the square root of 3. If in a system were also considered losses on the line, the resulting voltage values were less than the expected 230 V. In the system were considered large losses on lines, so the phase voltages at the terminals were lower, namely: Ua_A = 220.5 V, Ua_B = 221 V, Ua_C = 223.1 V, and Ua_D = 227 V. The demand current depends on the size of the load being connected at the consumption point (load). The active and reactive PQ load was unchanged in the circuit.
Table 2. Consumptions for simulation of steady-state off-grid network
.
Model of off-grid network with dynamic load
In this part of the simulation there was modified model of loads. Instead of the loads modeled by the constant value, dynamic loads were used that were controlled by external input. Dynamic load operation is described in the previous chapter definitions of loads During these simulations, two generators with a nominal power of 250 kVA and four loads A, B, C and D were used in which the phase voltage and current in phase L1 and power value in L1 were measured.
Fig.9. Network frequency response to output power
In Fig. 9, the consumed power is indicated by dynamic loads. Self-consumptions (2 x 12.5 kW) and parasitic loads to dynamic loads (3 x 9.5 kW + 4.75 kW), which are purely resistive, have to be added to the total output power. These parasitic loads are in the system because dynamic loads and synchronous generators cannot be in series with an inductive element of three-phase power lines. Those are described by the RL parameters listed in Table 3.
Table 3. Resistance and inductance of power lines in simulations with dynamic load
.
By a continual decreasing, respectively by increasing of the power consumption there was observed, that the regulators of the synchronous generators respond to these changes, and there is a decrease, respectively increase in output power produced by synchronous generators, but the frequency is not regulated to the nominal value of fn = 50 Hz. Thus, the frequency of the network will be short-lived at a different value near the nominal frequency due to the rate of decrease/increase of the consumed power. This can also be seen in Fig. 9, from 478 seconds to 595 seconds, the network’s frequency was around 50.2 Hz. From 900 s to 1000 s the network frequency was stable at values between 49.93 and 49.95 Hz.
Off-grid model with a wind turbine that operates at a dynamic wind speed
Wind simulation was used to simulate the wind-flow circuit as it is illustrated in Fig. 10. The wind loaded from the text file has a value of 9 m·s–1 at time t = 0, which is the nominal wind for the wind turbine used. Subsequently the wind varies around this value. Wind reaches a maximum value of 12 m·s–1. The wind turbine operates with winds ranging from 4.5 m·s–1 to 12.5 m·s–1.
Since the simulated wind turbine has no stabilizing mechanism, the supplied turbine power also varies around the nominal value. This was reflected negatively on network frequency and voltage. Since the simulated off-grid network is small in size, voltage fluctuations have been registered in all four A, B, C and D loads. Frequency of the grid and voltage at the wind turbine terminals are shown in Fig. 10. Referring to Fig 10 it can be seen that even with small wind changes, the frequency has risen above 51 Hz, or falls below 49 Hz. Voltage at wind turbine terminals is fluctuating. In case of a sudden change of wind, the voltage exceeds 250 V, respectively drops to 190 V. Since the voltage fluctuations are relatively strong, a digital flickermeter has been connected to load points A, B, C and D and to the wind turbine connection point.
Fig.10. Simulated wind parameters
Flicker-effect measurement in network with a wind turbine
In the previous section there was a description of offgrid operation with a wind turbine with dynamic wind. In order to determine the flicker effect in the aforementioned network, a digital flickermeter was added at points A, B, C and D to find a short-term flicker rate that is calculated at simulation time of 5 to 605 s, representing a ten-minute time period. In Table 4 is the measured short-term rate of flicker and averaged percentiles. The smallest rate of flicker shortterm perceptive was simulated with a constant wind velocity of 9 m·s–1 and a dynamic load. On the other side, the highest short-term flicker rate was simulated with dynamic wind speed and dynamic load. The short-term flicker rate was in accordance with standard STN EN 50160.
The Fig 11 shows the measured instantaneous level of the flicker effect at the load point C for dynamic load simulation (Fig. 9) and the dynamic wind simulation (Fig. 10). From Fig. 11, it is apparent that the blink effect was occurred in the case of dynamic load simulation at a time when the load was connected or disconnected in the network. It was observed for example, at time t1 = 100 s when a load of 30 kVA was disconnected at the load point B or at time t2 = 400 s when a load of 20 kVA was connected at the load point C or at time t3 = 800 s when the load of 20 kVA was disconnected from load point C. In the case when the dynamic wind acts on the wind turbine (see Fig. 10), the measured instantaneous level of flicker effect will appear as a stochastic noise.
Fig.11. Measured instantaneous level of flicker effect
The Table 4 shows the short-term flicker rate response for the load points A, B, C and D and for the point on the wind turbine terminals. The particularity of these results is that in each simulated scheme, at the load point A, the highest degree of short-term flicker is measured. This is due to the fact that the point A is powered by a line whose resistance and reactance is much larger than the lines connecting the other points (see Table 3). The voltage at point A in these simulations was stabilized at U = 189.6 V (in real conditions, such a low voltage would be a problem for the operation of many devices).
Table 4. Flicker effect in the simulated network
.
In order to reduce the influence of power line on the measured flicker effect, the simulations were repeated except that the line joining the load point A was simulated by resistance Rc = 0.0134 Ω and inductance Lc = 23.7 µH connecting the load point C. The results are given in Fig. 12.
Fig.12. Voltage in simulated scheme for point A
In Fig. 12 is the voltage characteristics at the load point A in the case where was considered constant wind of 9 m·s– 1 during the whole simulation and the dynamic load as described in section B. For the case 1 there was considered the original power line whose resistance was Ra = 0.2010 Ω and inductance La = 355.5 µH. In case 2, a point A was connected by a line with parameters Ra = 0.0134 Ω and La = 23.7 μH. From Fig. 12, it is clear that in case 1 there is a greater voltage fluctuation at the terminals at the load point A as in case 2. For example, during the disconnection of the 20 kVA load from the load point C, there was observed (in case 1) at the load point A the short-term voltage drop from Uf = 194.3 V to Uf = 149.5 V, which is a drop of ΔU = 44.8 V. In case 2, there was drop from Uf = 225.7 V to Uf = 199.4 V, which is drop of ΔU = 26.3 V. As there is less voltage fluctuation in transient phenomena, the resulting flicker effect will be less. In test example 2, the value of the Short Term Perceptibility (Pst) of flicker effect in point A was Pst = 0.335491 (the original value, in case A was Pst = 0.523317).
Conclusion
This article presented the particular results of off-grid network simulations with consideration of renewable resources (wind turbine) and without considering renewable energy sources. The simulated off-grid network consisted of two diesel generators with a nominal output of 250 kW and with loads A, B, C and D, representing 4 load points representing 4 off-grid sites. In the case of a load disconnection or connection, the generators are able to regulate the system so that the power output is equal to the power delivered. The regulation of diesel generators has ensured the control of the hydraulic turbine.
The problem of off-grid systems with a wind turbine (WT) is that the power output from the WT cannot be regulated. The WT produces electricity according to current climatic conditions. Therefore, in the case of rapid climate change, there is a rapid change in the output of the WT. For example, with decreased wind speed, a sudden drop in the electricity produced from the WT may occur. By adding a wind turbine into an off-grid, an increased flicker effect was observed. In addition to voltage fluctuations in the network, the network frequency also varies. Large frequency fluctuations can have a negative impact on diesel generators. Flicker effect occurs when disconnecting or connecting loads, resources, and off-grid networks. In both cases, it is necessary to consider how to remove the unfavorable phenomenon of blinking. For this reason, it is necessary to have good data for prediction of electricity production in island operation under the different wind generation modes.
Acknowledgement This work was supported by the Ministry of Education, Science, Research and Sport of the Slovak Republic and the Slovak Academy of Sciences under the contract No. VEGA 1/0372/18.
REFERENCES
[1] M. Špes, Ľ. Beňa, M. Kosterec, M. Márton, Determining the current capacity of transmission lines based on ambient conditions, In: Journal of Energy Technology. Vol. 10, no. 2 (2017), p. 61-69. ISSN 1855-5748 [2] R. Cimbala, L. Kruželák, S. Bucko, J. Kurimský, M. Kosterec, Influence of Electromagnetic Interference on Time-Domain Spectroscopy of Magnetic Nanofluids, In: EPE 2016. Danvers: IEEE, (2016), p. 279 – 282. ISBN 978-1-5090-0907-7 [3] D. Medveď, O. Hirka, Investigation of Electromagnetic Fields in Residential Areas. In: Acta Electrotechnica et Informatica. Vol.17, No. 3 (2017), p. 48-52. ISSN 1335-8243 [4] T. Košický, Ľ. Beňa, Optimizing deployment of battery storage systems, In: Current Problems of Maintenance of Electrical Equipment and Management. Košice: TU, 2014 p. 131-141. ISBN 978-80-553-1818-9 [5] V. Volokhin, I. Diahovchenko, V. Kurochkina, M. Kanálik, The influence of nonsinusoidal supply voltage on the amount of power consumption and electricity meter readings, In: Energetika. Vol. 63, no. 1 (2017), p. 1-7. ISSN 0235-7208 [6] Z. Čonka, M. Kolcun: Impact of TCSC on the Transient Stability In: Acta Electrotechnica et Informatica. Vol. 13, no. 2 (2013), p.50-54. – ISSN 1335-8243 [7] MathWorks, Model hydraulic turbine and proportional-integralderivative, https://www.mathworks.com/help/physmod/sps/powersys/ref/hydraulicturbineandgovernor.html [8] MathWorks, Three-Level Bridge, https://www.mathworks.com/help/physmod/sps/powersys/ref/threelevel bridge.html [9] MathWorks, Av. Model of a 100-kW Grid-Connected PV Array, https://www.mathworks.com/help/physmod/sps/examples/average-model-of-a-100-kw-grid-connected-pv-array.html [10] MathWorks, Implement three-phase dynamic load with active power and reactive power as function of voltage or controlled from external input, https://www.mathworks.com/help/physmod/sps/powersys/ ref/threephasedynamicload.html [11] Ž. Eleschová, A. Beláň, B. Cintula, B. Bendík, Smart grids analysis – View of the transmission systems voltage stability, In EPE 2018. Brno: University of Technology, 2018, p. 37-42. ISBN 978-1-5386-4612-0 [12] D. Kaprál, P. Braciník, M. Roch, M. Höger, Optimization of distribution network operation based on data from smart metering systems, Electrical Engineering, Vol. 99, Issue 4, Springer, New York, USA, 2017, December, pp: 1417-1428, ISSN 0948-7921
Authors: Ing. Dušan Medveď, PhD. Technical University of Košice, Department of Electric Power Engineering, Mäsiarska 74, 04001 Košice, Slovakia E-mail: dusan.medved@tuke.sk; Ing. Zsolt Čonka, PhD. Technical University of Košice, Department of Electric Power Engineering, Mäsiarska 74, 04001 Košice, Slovakia E-mail: zsolt.conka@tuke.sk; Ing. Marek Pavlík, PhD. Technical University of Košice, Department of Electric Power Engineering, Mäsiarska 74, 04001 Košice, Slovakia E-mail: marek.pavlik@tuke.sk; Ing. Ján Zbojovský, PhD. Technical University of Košice, Department of Electric Power Engineering, Mäsiarska 74, 04001 Košice, Slovakia E-mail: jan.zbojovsky@tuke.sk; Dr.h.c. prof. Ing. Michal Kolcun, PhD. Technical University of Košice, Department of Electric Power Engineering, Mäsiarska 74, 04001 Košice, Slovakia E-mail: michal.kolcun@tuke.sk; Ing. Michal Ivančák, Technical University of Košice, Department of Electric Power Engineering, Mäsiarska 74, 04001 Košice, Slovakia E-mail: michal.ivancak@tuke.sk
Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 95 NR 7/2019. doi:10.15199/48.2019.07.31
Published by Grzegorz MASŁOWSKI, Robert ZIEMBA, Tomasz KOSSOWSKI, Rzeszow University of Technology
Abstract. The results of simulations of overvoltages induced in the overhead transmission line, caused by nearby lightning stroke are presented. Calculations were made using the LIOV module implemented in the EMTP-RV program. The influence of the distance of the lightning channel from the line to the overvoltages has been investigated. The results for the various lightning currents in the lightning channel have been compared.
Streszczenie. Przedstawiono wyniki symulacji przepięć indukowanych w linii napowietrznej od pobliskich wyładowań atmosferycznych. Obliczenia wykonano przy użyciu modułu LIOV zaimplementowanego w programie EMTP-RV. Zbadano wpływ odległości kanału pioruna od linii oraz wpływ kształtu prądu piorunowego na kształt i wartości szczytowe indukowanych przepięć. (Przepięcia w liniach napowietrznych wywołane pobliskim wyładowaniem atmosferycznym).
Keywords: lightning protection, lightning induced overvoltages, overhead power line. Słowa kluczowe: ochrona odgromowa, indukowane przepięcia piorunowe, napowietrzna linia energetyczna.
Introduction
Lightning-induced overvoltages are transient overvoltages on overhead power lines caused by indirect lightning events, i.e. lightning strikes hitting the ground or objects in the vicinity of the lines. In accordance with lightning protection standards IEC 62305 [1-3], the impact of nearby lightning strikes shall be taken into account in the design of the Lightning Protection System (LPS). Although the effects of nearby lightning strikes are smaller than direct impacts, the range of impact is much greater. They cause overvoltage in the indoor installations and in the lines coming into the building.
Computer code LIOV (Lightning Induced Overvoltages) [4-6] was implemented as the special module in EMTP-RV software. Now LIOV module allows for the calculation of lightning-induced voltages along a multiconductor overhead line as a function of:
– lightning current waveshape (peak value, front steepness, and duration), stroke location and return stroke velocity; – line geometry (height, length, number and position of conductors) and line terminations; – ground resistivity and relative permittivity.
The geometry for the calculation of LEMP (Lightning Electromagnetic Pulse) and its coupling with an overhead line, implemented in LIOV module is showed in Fig. 1.
Fig. 1. Geometry for the calculation of Lightning Electromagnetic Pulse (LEMP) and its coupling with an overhead line, implemented in LIOV code.
LIOV code adopts an engineering return stroke model and the lightning channel is assumed as a straight vertical antenna [7]. An engineering return stroke model is a formula that describes the spatial and temporal distribution of the return stroke current along the lightning channel, as a function of the current waveshape at the base of the channel and one or two additional parameters. This model with the straight and perpendicular lightning channel was simulated in [8-10] with the MTLL and the MTLE return stroke models in frequency domain.
In the present version of LIOV-EMTP module only the Transmission Line (TL) return stroke model is adopted [11, 12].
Modelling and simulation of lightning induced overvoltages
The most commonly adopted return-stroke models to calculate lightning-induced voltages are [10]: the Modified Transmission Line Linear (MTLL) model; the Modified Transmission Line Exponential (MTLE) model; and the Transmission Line (TL) model. In the MTLL model the current wave propagates without distortion but its peak value decays linearly with height.
In the MTLE model the current wave propagates similarly but the current peak value decays exponentially with height.
In the TL model, it is assumed that the current wave at the base channel propagates up the lightning channel without distortion and without attenuation, at a constant speed v. The return stroke current in the lightning channel at height z is:
.
where: 1(t) is the Heaviside function equal to 1 for t ≥ z/v and zero otherwise; v is the upward propagating return stroke velocity; i0(t) is the current at the base of the channel.
The return stroke velocity is expected to be between 100 and 200 m/μs [13]. The channel base current waveform can be represented by means of Heidler function, which is a function of the lightning current for analysis purposes in IEC 62305-1 standard, specified by equation:
.
In (2) I is the peak current, k is the correction factor for the peak current, τ1 is the front time constant and τ2 is the tail time constant.
According to IEC 62305-1 [1], the current shapes of the first positive impulse 10/350 µs, the first negative impulse 1/200 µs and the subsequent negative impulses 0.25/100 µs, defined by Heidler function, were taken for calculation. The peak values for the first lightning protection level (LPL) were chosen (respectively 200 kA, 100 kA and 50 kA). Comparison of waveshape of the lightning current components is shown in Fig. 2.
Fig.2. Waveforms of the lightingThe channel base current (2) can be set in the “liov options” device as shown in Fig. 3, assuming peak value I02 equal zero. current components.
Fig.3. Definition of current waveform in LIOV option module.
In lightning protection standard [2] there are defined the collection areas of flashes directly to line (Al) and to ground near line (AL). This collection areas are taking into account for calculation of the risks of lightning losses. The width of these areas are different in two subsequent editions of IEC 62305-2 standard. The width of collection area of flashes into the line was increased from WI = 30 m to WI = 40 m and the width of the collection area of indirect lightning flashes was increased from WI = 1000 m to WI = 4000 m. As we can see, in the case of indirect discharges, the collection area has been increased four times. This increase of the collection area can be of importance in assessing the risk of the lightning damage in the design of the LPS.
The example of a system that can be simulated with LIOV-EMTP module is shown in Fig. 4. This case study was simulated by using the circuit shown in Fig. 5. The overhead three-wire transmission line of 1000 m in length and 10 m in height was taken to the simulations. Position of the lightning channel is xs = 500 m and ys (distance of the lightning channel from power line is variable: ys1 = 100 m, ys2 = 500 m, ys3 = 2000 m).
The geometrical data of the conductors of the overhead line are represented in Fig.6, where the subscripts of H, D and d refer to the pins corresponding to the line ends.
Fig. 4. Scheme of simulated system.
Fig.5. Analysed circuit in EMTP-RV software.
Fig.6. Definition of conductor geometry in transmission line.
The ends of the transmission line are terminated by “line match” components containing connected to ground characteristic impedances of the line.
Results of the simulation
In Fig. 7 and Fig. 8 are shown the induced overvoltages at the ends of transmission lines, for different lightning current impulses and for different soil parameters. In Fig. 7 are shown results of simulations for case study with lossy ground (soil resistivity: ρ = 100 Ωm). Strike distance ys is variable: 100 m, 500 m and 2000 m. Simulations has been conducted for the first positive impulse 10/350 μs with peak value 200 kA; the first negative impulse 1/200 μs with peak value 100 kA; and for the subsequent negative impulse 0.25/100 μs with peak value I = 50 kA. In Fig. 8 are shown results of simulations for case study with lossless ground, and for soil with resistivity ρ = 100 Ωm and ρ = 500 Ωm. Strike distance ys = 500 m. Simulations has been conducted for the first positive impulse 10/350 μs with peak value I = 200 kA; the first negative impulse 1/200 μs with peak value I = 100 kA; and for the subsequent negative impulses 0.25/100 μs with peak value I = 50 kA.
.
Fig. 7. Comparison of the induced overvoltages for case study with lossy ground (soil resistivity: ρ = 100 Ωm). Strike distance ys is variable: 100 m, 500 m and 2000 m: a) the first positive impulse 10/350 μs, peak value 200 kA; b) the first negative impulse 1/200 μs, peak value 150 kA; c) the subsequent negative impulses 0.25/100 μs, peak value 100 kA.
.
Fig. 8. Comparison of the induced overvoltages for case study with lossless ground and for soil resistivity ρ = 100 Ωm and ρ = 500 Ωm. Strike distans ys = 500 m. a) the first positive impulse 10/350 μs, peak value 200 kA; b) the first negative impulse 1/200 μs, peak value 100 kA; c) the subsequent negative impulses 0.25/100 μs, peak value 50 kA.
To compare voltage distribution along the line in Fig. 9 were shown overvoltages in the middle (xs = 500 m) and at the end of the line (xs= 0 m) for different distance of the lightning channel from the line. Simulation was for the first positive impulse 10/350 μs with amplitude 200 kA.
.
Fig. 9. Comparison of the induced overvoltages in the middle and at the end of the line for first lightning stroke 10/350 200 kA, for different distance of the lightning channel from the line: a) ys = 100 m, b) ys = 500 m, c) ys = 2000 m
Conclusions
The paper presents results of the computer simulations of the overvoltages induced in the transmission line by nearby lightning stroke. Influence of the striking distance, soil resistivity and shape of the lightning current were investigated. As shown in Fig. 7 maximum values of induced overvoltages strongly depend on the distance of the lightning channel from the line. Results in Fig. 8 shows that soil resistivity does not have a significant impact on maximum values and shapes of induced overvoltages within the considered distances. As shown in Fig. 9, the peak values of the induced voltages at the centre of the line relative to the overvoltages on the ends of the line depend on the distance of the lightning channel from the line.
REFERENCES
[1] IEC 62305-1:2010 Protection against lightning – Part 1: General principles. [2] IEC 62305-2:2010 Protection against lightning – Part 2: Risk management. [3] NFPA 780 Standard for the Installation of Lightning Protection Systems. 2014. [4] C. A. Nucci and F. Rachidi, “Interaction of electromagnetic fields with electrical networks generated by lightning,” in The Lightning Flash: Physical and Engineering Aspects, V. Cooray, Ed. IEE – Power and Energy Series 34, 2003, pp. 425–478. [5] F. Napolitano, A. Borghetti, C. A. Nucci, M. Paolone, F. Rachidi, and J. Mahserejian, “An advanced interface between the LIOV code and the EMTP-RV,” presented at the 29th Int. Conf. Lightning Protection (ICLP), Uppsala, Sweden, 2008 [6] M. Paolone, F. Rachidi, A. Borghetti, C. A. Nucci, M. Rubinstein, V. A. Rakov, and M. A. Uman, “Lightning electromagnetic field coupling to overhead lines: theory, numerical simulations, and experimental validation”, IEEE Trans. Electromagn. Compat., vol. 51, no. 3, pp. 532–547, 2009 [7] Masłowski G., Rakov V.A.: New Insights Into Lightning Return-Stroke Models with Specified Longitudinal Current Distribution. IEEE Trans. Electromagn. Compat., Vol. 51, No. 3, August 2009, 471–478 [8] Masłowski G. Ziemba R.: Calculation of lightning-induced voltages inside the structure using engineering return-stroke models. Proc. 28th International Conference on Lightning Protection, Kanazawa, Japan, 2006, 1132-1137 [9] Masłowski G., Ziemba R.: Modelowanie przepięć atmosferycznych w liniach elektroenergetycznych z uwzględnieniem kanału pioruna. Przegląd Elektrotechniczny, 3/2007, 153–156. [10] Masłowski G.: Współczesne trendy w modelowaniu wyładowań atmosferycznych – teoria i zastosowania, Przegląd Elektrotechniczny (Electrical Review), R. 86 NR 11a/2010, 308–312 [11] M. A. Uman and D. K. Mclain, “Magnetic field of lightning return stroke,” J. Geophys. Res., vol. 74, no.28, pp. 6899–6910, 1969. [12] F. Napolitano, “An analytical formulation of the electromagnetic field generated by lightning return strokes,” IEEE Trans. Electromagn. Compat., vol. 53, no. 1, pp. 108–113, 2011. [13] V. A. Rakov and M. A. Uman, Lightning: Physics and Effects. Cambridge, 2003.
Authors dr hab. inż. Grzegorz Masłowski, Politechnika Rzeszowska, Wydział Elektrotechniki i Podstaw Informatyki, ul. Powstańców Warszawy 12, 35-959 Rzeszów, e-mail: maslowski@prz.edu.pl; dr inż. Robert Ziemba, Politechnika Rzeszowska, Wydział Elektrotechniki i Podstaw Informatyki, ul. Powstańców Warszawy 12, 35-959 Rzeszów, e-mail: maslowski@prz.edu.pl; mgr inż. Tomasz Kossowski, Politechnika Rzeszowska, Wydział Elektrotechniki i Podstaw Informatyki, ul. Powstańców Warszawy 12, 35-959 Rzeszów, e-mail: t.kossowski@prz.edu.pl;.
Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 94 NR 2/2018. doi:10.15199/48.2018.02.10
Published byDániel MARCSA, eCon Engineering Kft., Hungary
Abstract. Transformer noise is a significant contribution to unwanted ambient noise, especially in the vicinity of the electrical transmission facility. It is therefore very important to get to know the mechanism of noise generation of the distribution transformer. As outcomes of this work, a finite element based multiphysics model is presented which provides a convenient and efficient toolchain for simulating the transformer sound emission mechanism. Finally, the operation of modelling chain is presented on a 200kVA distribution transformer simulation.
Streszczenie. Hałas transformatora ma znaczący wpływ na niepożądany hałas otoczenia, zwłaszcza w pobliżu instalacji przesyłowej prądu elektrycznego. Z tego powodu ważnym jest poznanie mechanizmu generowania szumu transformatora rozdzielczego. Jako wynik tej pracy przedstawiono model transformatora rozdzielczego 200 kVA oparty na analizie elementów skończonych, który zapewnia wygodny i wydajny zestaw narzędzi do symulacji mechanizmu emisji dźwięku z analizowanego urządzenia. (Analiza hałasu i wibracji transformatora rozdzielczego).
Keywords: Noise and vibration, finite element analysis, coupled simulation, distribution transformer, ANSYS. Słowa kluczowe: hałas i wibracje, analiza elementów skończonych, symulacja sprzężona, transformator dystrybucyjny, ANSYS.
Introduction
Distribution transformers are one of the most critical components for electrical energy transportation and distribution. The vibration and noise of these electric machines increasingly interested designers and manufacturers. It is therefore critical that manufacturers are able to accurately identify the acoustic characteristics of a transformer before production commences. However, the various physical phenomena are strongly related in the transformers, as illustrated in Fig. 1, so only the multiphysics or coupled numerical simulation can be useful to get knowledge about these effects. Further, safety regulations require that the noise level is kept within a certain range.
The study of noise and vibration in transformers began in the 1930s, mainly by transformer manufacturers [1]. These works are focused on mainly the measurement of transformer noise. In recent years, thanks to the computer capabilities and software, more and more attention has been paid to the numerical simulation of these unwanted effects. However, most of these works are separated the strongly coupled sources or effects of noise and vibration. The noise mainly originates from the magnetostrictive effect of the steel sheet [2], [3] and the shape of the core [4]. The electromagnetic force produced in the windings also important as electromagnetic noise source [5]. The clamping stress and natural frequencies of core and tank also have some effect on vibration [6]. These effects result in deformation of the tank [7], [8], which cause disturbing audible sound. When using coupled or multiphysics simulation, important effects are neglected or analysed only a special load case. Most of the time, the permeability of the transformer core is isotropic [5], [9] or analysed the shortcircuit state of transformer [5], [10]. But a coupled simulation workflow with the whole noise generation process at nominal operation has not been studied.
This work focuses on distribution transformer coupled simulation with the help of the finite element method (FEM) [10], [11]. The aim of this paper is to develop a numerical methodology that accurately predicts the vibration and acoustic characteristic of a distribution transformer under normal operation condition. The modelling procedure is based on the chaining of three analysis methods, the electromagnetic, the mechanical and the acoustic simulations as you can see in Figure 1. The weak or series coupling was used for the connection because in this case completely identical geometry is not needed, finite element mesh and solver options are independent for each analysis.
Fig.1. The process of noise generation by the transformer.
The workflow of the transformer simulation was established using ANSYS Workbench environment. The auxiliary noise sources are not taken into account.
Transformer Coupled Simulation
The geometry of the 200kVA distribution transformer with the sectioned tank is shown in Fig 2a. The transformer core is composed of oriented silicon steel sheets, and the anisotropy of silicon steel sheets should be considered. The B-H curve and magnetostrictive curve of steel sheet in rolling and transverse direction are shown in Fig 3. The key parameters of the transformer are summarized in Table 1.
Table 1. Main technical parameters of the analysed oil-type transformer.
.
Fig.2. CAD model of the distribution transformer (upper) and magnetic flux density vectors in the laminated core (lower).
Fig.3. Magnetisation curves (left) and magnetostriction curves (right) of steel sheet in rolling and transverse direction.
The simulation chain starts with time-dependent nonlinear electromagnetic simulation. The external circuit of excitation and load is directly coupled to the finite element model. The geometry (the active part of the transformer) of the problem is discretized into 273823 tetrahedron elements. The numerical solution of these equations is based on ungauged T, Φ – Φ – formulation [11], [12],
.
where [σ] and [μ] is the conductivity and permeability tensor, T0 is the impressed current vector potential, T is the current vector potential, and HP is the additional field component due to core loss.
The main task of this step is to calculate the Maxwell force and the Lorentz force in the core and windings, respectively. In addition, the different power losses (stranded, eddy current, hysteresis) in the core and the winding can also be determined in this step. The visualization of the magnetic flux density vectors in the core can be seen in Fig. 2b.
The next step is mechanical simulation in the frequency domain. However, resonance in the transformers may be induced when the multiples of excitation frequency are sufficiently close to the natural frequency [6], therefore it should be carried out the modal analysis. The main origin of vibration is the magnetostrictive strain, so the most critical frequency is 100 Hz. The modal analysis results are shown in Fig. 4, where it can be seen that both of them one frequency cannot avoid the 100 Hz.
Fig.4. Simulated results of the natural frequency. The active part vibration at 104.4 Hz (upper) and the tank vibration at 98.8 Hz (lower) natural frequency.
Using the results of the electromagnetic simulation (magnetostriction, Lorentz force) and the natural frequencies from the modal analysis the mechanical displacement is evaluated by harmonic analysis. The solved generalized equation of motion is given as [4]
.
where Mu is the structural mass matrix, Cu is the viscous damping matrix, Ku is the stiffness matrix, ü ,u̇ , u is the nodal acceleration, nodal velocity and nodal displacement vector, respectively. fe is the spectrum of force from the electromagnetic simulation as load force.
The basic procedure to pass the electromagnetic results from the time domain to the frequency domain mechanical simulation is the Fourier transform [10] of results. The vibration from the core due to magnetostriction contains a 100Hz component (twice the frequency of power source frequency) and harmonics, while the vibration from the winding has mainly a pure 100 Hz tone if the current in the winding themselves are free of harmonics [13].
Figure 5 shows the total deformation of the active parts of the transformer at 100 Hz. The deformation of core and clamp can be seen in Fig 5a and the deformation of the primary and secondary winding in Fig 5b. The maximal deformation at the core is in the upper yoke, where the displacement is greater than 3.5 μm. The maximum of displacement in the windings is right one (Phase C). The maximum deformation of this winding is 3.1 μm.
Finally, using the harmonics displacement, we determine the resulting pressure level of acoustic waves propagation through the insulation oil, the tank and the surrounding air. The nodal velocities of active part from the harmonic analysis have been interpolated and mapped to the acoustic mesh of oil. Fig. 6 illustrates the velocity vectors on the oil inner surface. The numerical prediction of sound radiation has required the oscillation of the transformer tank. Therefore, it is necessary the coupling of (3) and the Navier-Stokes equation of fluid momentum and the flow continuity equation,
.
Fig.5. The deformation of the active part. The core and clamp deformation (upper) and the deformation of winding (lower).
Fig.6. The nodal velocity vectors mapped to the acoustic body surrounding the transformer tank.
Fig.7. The x-, y- and z-component of acceleration on the surface of the middle limb of core.
Fig.8. The x-, y- and z-component of acceleration on top of the transformer tank.
where ρ0 is the mean fluid density, Mq is the fluid mass matrix, Cq is the fluid damping matrix, Kq is the fluid stiffness matrix, fq and fu is the load force and Cfs is the fluid-structure coupling term. The pressure is p = q̇ = jωq.
Results and Discussion
The main sources of vibration are the electromagnetic origin, so the accuracy of the electromagnetic model is important. The loss has been used to validate the model. The calculated total loss is 2076 W, which corresponds to the value specified in the datasheet (see in Table I). The main reason for the difference is that there is no information on the load used for the measurement, and an average distribution network as load used in the finite element simulation.
Fig. 7 and 8 show the results of mechanical harmonic analysis at two specific points on the transformer. These figures show the spectrum of x-, y- and z-component of the acceleration. As the modal analysis has shown, one of the resonance frequencies of the active part and the tank is close to twice the excitation frequency. This is also supported by Fig. 7 and 8, because one of the peak values of the acceleration spectrum is at 100 Hz. The tank top has another peak in the spectrum at 150 Hz. Based on this information, the sound pressure level is analysed at 100 Hz. Fig. 9 and 10 summarize the simulation results from the acoustic field simulation. These figures show the sound pressure level at 2 m from the tank wall. 0 degrees and – 180 degrees indicate the centre of the shorter side of the transformer. Reference line shows the 53 dB, which is the noise level of this transformer based on datasheet. These results also support that the accuracy of the result obtained by the numerical simulation is acceptable. As shown in the figures, the sound pressure reaches 60 dB in the 680 mm case. When using A-weighting [13], the sound pressure level increases as shown in Fig. 10. Based on the results, it can be stated that the analysed transformer meets the requirements, but its noise level can be reduced by proper design.
Fig.9. Sound pressure level [dB] around the transformer at 340 mm and 680 mm height and 2000 mm distance from tank wall.
Fig.10. Sound pressure level [dB] and its A-weighted version around the transformer at 680 mm height and 2000 mm distance from tank wall.
Conclusions
This paper analyzes the performance of noise and vibration in the distribution transformer considering the anisotropy and magnetostriction influence of silicon steel sheet. A three-dimensional finite element method based multiphysics workflow in the electromagnetic – mechanical – acoustic field is established using sequential coupling of ANSYS software. The operation of the 3-D finite element workflow is analyzed via a 200 kVA distribution transformer problem. It shows that the simulation is in a reasonable agreement with the transformer datasheet value, verifying the validity of the presented coupled simulation. The presented simulation workflow seems to be appropriate for simulating transformer noise and vibration or it may be helpful to develop new transformer diagnosing method.
The future plan is to further develop the presented workflow to take into account the structure-borne transmission of sound waves through the transformer mountings and auxiliary noise sources, e.g. oil pump. In addition, the speed up and simplification of simulation workflow also an ongoing task.
The research for this paper was financially supported by the EU and the Hungarian Government from the project “Intensification of the activities of HU-MATHS-IN – Hungarian Service Network of Mathematics for Industry and Innovation” under grant number EFOP-3.6.2-16-2017- 00015.
REFERENCES
[1] IEEE Commi t te Repor t , Bibliography on Transformer Noise, IEEE Transactions on Power Apparatus and Systems, PAS-87 (1968), 372-387 [2] Zhang P., Li L., Cheng Z., Tian C., Han Y., Study on Vibration of Iron Core of Transformer and Reactor Based on Maxwell Stress and Anisotropic Magnetostriction, IEEE Transactions on Magnetics, 55 (2019), No. 2, 9400205 [3] Chen D., Hou B., Feng Z., Bai B., Study of Magnetostrictive Influence of Electrical Sheet Steel Under Different DC Biases, IEEE Transactions on Magnetics, 55 (2019), No. 2, 2001305 [4] Shuai P., B iela J., Impact of Core Shape and Material on the Acoustic Noise Emission of Medium Frequency, Medium Voltage Transformers, 17th European Conference on Power Electronics and Applications (EPE’15 ECCE-Europe), Geneva (2015), 1-11 [5] Duan X., Zhao T., Liu J., Zhang L., Zou L., Analysis of Winding Vibration Characteristics of Power Transformers Based on the Finite Element Method, Energies, 11 (2018), No. 9, 2404 [6] Hsu C.-H., Lee S.-L., Lin C.C., Liu C.-S., Chang S.-Y., Hsieh M.-F., Huanf Y.-M., Fu C.-M., Reduction of Vibration and Sound-Level for a Single-Phase Power Transformer with Large Capacity, IEEE Transactions on Magnetics, 51 (2015), No. 11, 8403204 [7] Vieira N., Antunes P.J., Martins C., Dias G.R., Coelho A.T., Vibro-Acoustic Analysis of a Distribution Power Transformer Using the Finite Element Method, CWIEME 2008 – Coil Winding, Insulations & Electrical Manufacturing, Berlin (2008), 1-10 [8] Shengchang J., Lingyu Z., Yanming L., Study on Transformer Tank Vibration Characteristics in the Field and Its Application, Przegląd Elektrotechniczny, 2011 (2011), No. 2, 205-211 [9] Kubiak W., Wi tc zak P., Vibration Analysis of Small Power Transformer, COMPEL – The International Journal for Computation and Mathematics in Electrical and Electronic Engineering, 29 (2010), No. 4, 1116-1124 [10] Kaltenbacher M., Numerical Simulation of Mechatronic Sensors and Actuators, Springer-Verlag, Berlin, 2007 [11] Kuczmann M., I ványi A., The Finite Element Method in Magnetics, Akadémiai Kiadó, Budapest, 2008 [12] Lin D., Zhou P., Chen Q.M., Lambert N., Cendes Z.J., The Effects of Steel Lamination Core Losses on 3D Transient Magnetic Fields, IEEE Transactions on Magnetics, 46 (2010), No. 8, 3539-3542 [13] Tímár P.L., Fazekas A., Kiss J., Miklós A., Yang S.J., Noise and Vibration of Electrical Machines, Akadémiai Kiadó, Budapest, 1989
Author: dr. Dániel Marcsa, Ph.D., eCon Engineering Kft., Kondorosi u. 3, Budapest, H-1116, Hungary, E-mail: daniel.marcsa@econengineering.com.
Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 95 NR 12/2019. doi:10.15199/48.2019.12.38
Published by Harsha Korde, EE Power – News: The Impact of EVs on the Electric Grid, May 14, 2022
With gas prices rising, electric vehicles are becoming more popular, but transportation decarbonization remains a major issue.
With gasoline prices rising and the effects of climate change increasingly acute, transportation electrification is gaining steam in countries throughout the world. The benefits of a transition toward electric vehicles (EVs), and away from the compressed natural gas (CNG) fueling those with traditional internal combustion engines (ICEs), are many: reduced carbon emissions, less noise pollution, improved air quality and enhanced energy efficiency, among others.
Planning for Electric Vehicles
That said, transportation decarbonization remains a significant issue in many countries whose power systems are dominated by fossil fuels. Though the switch to EVs will have direct repercussions for power grids, power system planning to assess EV impacts is practically nonexistent, leaving systems managers with an incomplete view of the technical and economic impact of EV integration.
Electric vehicles (EVs) charging. Image used courtesy of Pixabay
As EVs increasingly supplant ICE-based vehicles, changing demands in power utilization will play an ever more critical role. These demands can have a range of impacts on the power network, such as an increase in the number of short circuit currents, and can lead to voltage level violations, in addition to affecting electrical equipment such as transformers.
Chief among concerns is the load placed on the grid by EV charging. In a world where millions of EVs saturate the road, at any moment the grid could face an influx of stress from simultaneous, mass charging. That uncertainty makes it significantly harder for operators to balance grid supply and demand in real-time. At scale, an increased load on the distribution network impacts its power quality, and if EV batteries are charged without an analysis of their impact on the distribution network, it may directly result in an increase of energy unserved by the power system or the need for additional peak load capacity.
The Distribution Level
To address this issue at the distribution level, different load management schemes should be implemented alongside existing distribution network policy, including time-varying tariffs and incentives for different charging behaviors. But even with these, the frequent connecting and disconnecting of high-current EV batteries pose its own challenges to the efficient operation of the electrical power system.
The net electrical energy utilized by EVs in a specific area is termed a “charging load curve” of EVs for that area. To analyze the impact of EV penetration on the electrical grid, the predictions provided by this load curve are essential. In impact evaluation, an EV load curve analysis can aid in evaluating various fundamental parameters of the electrical power distribution system, such as overloading, the impact on a domestic transformer, power loss in the system, the stability of the grid, fluctuations in voltage, power quality, and stress on distribution cables or conductors, and so on. Developing a full picture of those impacts as a result of EV charging is vital, as the prime objective here is to construct charging infrastructure integrated with the grid well enough to ensure smooth operation and maintenance of the distribution network.
When EV charging is carried out in a three-phase power system, it results in voltage imbalance, since those chargers are single-phase. Because of the increased load, the introduction of power electronics to the charging process also results in the injection of harmonics. This can be the cause of rising transformer temperatures at the distribution feeder, leading to wear of the transformer bushings. Harmonic distortion can as well affect the interruption capability of circuit breakers. These issues can be res
V2G Technology
EV smart charging involves vehicle-to-grid (V2G) integration technology, allowing car batteries to give back to the power grid. In this way, the high-capacity batteries powering EVs can function as backup storage for the electrical grid. This type of setup utilizes bidirectional charging stations, where the power flow is directional based on the electricity demand at any given time. The extra energy can be used to power houses, buildings or anything connected to the power grid.
V2G integration technology has numerous benefits, such as improving the efficiency of power distribution. In a scenario where most EVs are charged simultaneously during peak hours, or at any time when energy demand is high, the system could easily be overloaded. With V2G technology, though, power companies can expand their capacity to meet these peaks; the bidirectional energy flow of V2G offers the most efficient model of power distribution. An additional benefit of this technology is the increase it could promote in renewable energy utilization, such as solar and wind, which will play an important role in sustaining the economy. Though these sources may be inconsistent, an efficient power grid can capture energy through them whenever needed and store it for distribution. Still, whenever there is a surge in energy capture, perhaps thanks to high winds, grid-level system storage has the potential to be maxed out. And that’s where EV batteries and V2G technology come in—with millions of EVs on hand ready to charge, there is additional room to capture and utilize this extra energy. When taken together, the above benefits also lead to another: cost stability. The more strain on the system, the higher the costs. Given that, the improved balance between energy supply and demand will naturally lead to less volatile pricing.
An Uphill Battle?
All that said, there is still much work to be done in fully realizing bidirectional V2G integration technology. Today, most electric vehicles and charging stations are unidirectional, and converting them to the bidirectional form will require significant investment. There is also no standardized cadre of rules and regulations governing V2G integration technology, but rather a hodgepodge of electrical standards applied across varying jurisdictions, making the implementation of such technology difficult. In the face of these challenges, there is a lack of clear incentives for household and business customers to convert to smart charging systems. As such, in moving forward with an effective transition to true V2G integration technology infrastructure, a top priority is to address these obstacles.
Power Quality Blog 