Power Quality Improvement in Distributed Generation System under varying Load Conditions using PWM and Hysteresis Controller

Published Vijayshree G1, Sumathi S2, RNSIT, Bangalore,Karnataka,India(1). RNSIT, Bangalore,Karnataka, India(2)

Abstract – This paper proposes a power system network modeled using a microgrid, integrated with wind and solar photovoltaic (PV) resources, along with the battery energy storage system (BESS) connected to the three-phase grid feeding the linear and nonlinear load. The simulation is carried out with unit vector and instantaneous reactive power control algorithm for series and shunt active power filter respectively. The power quality improvement is analyzed for the voltage sag, swell, and harmonics for system with load variation and without load variation. The performance analysis is based on voltage & current THD, voltage & current RMS, and power and power factor analysis. The THD for comparison is represented in form of bar chart to show the effective performance of control algorithm proposed using PWM method and hysteresis controller.

Streszczenie. W artykule zaproponowano modelowanie sieci elektroenergetycznej za pomocą mikrosieci zintegrowanej z wiatrowymi i słonecznymi źródłami fotowoltaicznymi (PV) wraz z bateryjnym systemem magazynowania energii (BESS) podłączonym do sieci trójfazowej zasilającej obciążenie liniowe i nieliniowe. Symulacja prowadzona jest za pomocą wektora jednostkowego i algorytmu regulacji mocy biernej chwilowej odpowiednio dla filtru mocy czynnej szeregowego i bocznikowego. Poprawa jakości energii jest analizowana pod kątem zapadów, wzrostów napięcia i harmonicznych dla systemu ze zmiennym obciążeniem i bez zmiennego obciążenia. Analiza wydajności opiera się na THD napięcia i prądu, wartości skutecznej napięcia i prądu oraz analizie mocy i współczynnika mocy. THD dla porównania przedstawiono w postaci wykresu słupkowego, aby pokazać efektywne działanie algorytmu sterowania zaproponowanego przy użyciu metody PWM i regulatora histerezy. (Poprawa jakości energii elektrycznej w systemie generacji rozproszonej w warunkach zmiennego obciążenia z wykorzystaniem PWM i regulatora histerezy)

Keywords: microgrid, distributed generation, power quality, UPQC
Słowa kluczowe: mikrosieci, rozproszone źróda energii, PWM

Introduction

Power quality (PQ) is used to assess and maintain the good quality of power at different levels (generation, transmission, distribution, and utilization) of the AC electrical power system. Power quality has become a challenging area of research in electrical engineering. Power quality is a combination of current and voltage quality. The voltage or current deviation from the ideal value is the disturbance of power quality. However, distinguishing between the voltage and current disturbances in the power system is difficult because for different customers different event leads to different disturbances. Therefore, in general, power quality is related to disturbances in voltage, current, frequency, and power factor. The electrical network considered for power quality assessment includes a three-phase programmable source connected to a microgrid with wind and solar PV resources along with battery storage. The system is connected to the non-linear load and linear load for power quality analysis. In [1] author explains generalized integrator controller which provides unity power factor (UPF) operation, load balancing, harmonics mitigation, and reactive power compensation for three phases single-stage grid interfaced solar energy conversion system. Different methods of analysis that describes the harmonic behavior of the electrical network and these data are reported to provide suitable mitigation strategies in a real-time operation is explained in [2]. The author in [3] explains the implementation of distributed generation operated for nonlinear and unbalanced load with different source conditions using enhanced Instantaneous Power Theory. The use of renewable energy sources has different environmental benefits and is also economical when compared with the traditional source of power generation [4]. The integration and control of renewable energy in electric power systems is mathematically modelled in [5] .In [6] ,a modified p-q theory-based control that implements a solar photovoltaic (PV) array integrated unified power quality conditioner (PV-UPQC-S) for analysis of the steady and dynamic performance of the system is explained. The paper [7] proposes a three-phase transformer less Hybrid Series Active Filter (THSeAF) in combination with SRF and PQ theory to provide suitable mitigation for power quality issues. [8] explains about the stationary and non-stationary power quality disturbance which is mitigated using variational mode decomposition (VMD) and decision treebased detection method which is capable of accurate detection, estimation, localization, and classification of all kinds of PQ disturbances in both noisy and noise-free cases. [9] explains how fluctuation in photovoltaic (PV) power plants affects the power quality and stability of the grid along with the increasing penetration of PVs. The addition of an energy storage system resolves this issue. The importance of battery parameter state of charge is taken care of to ensure the stability of the energy storage system. In [10], different positions of UPQC placement for better performance are analyzed. A Synchronous reluctance generator is used as an input to feed linear and nonlinear loads with an adaptive neural network-based control algorithm to improve power quality which mitigates some power quality problems such as harmonic suppression, and load balancing, reduction in frequency variation and voltage regulation[11]. Passive and active filters in providing power quality compensation, the advantages and disadvantages of both types of filters are explained and justify that the active power filter is more advantageous [12].

Problem formulation

The three-phase voltage source of 415V, 50 Hz is integrated with a Microgrid designed with solar and wind generation connected to linear and nonlinear load . The magnitude of the voltage at the microgrid is maintained the same as the supply voltage as it is connected in parallel. The supply voltage is observed to be 229.2 volts phase to phase as shown in figure 1, at t=0s to t=0.1s. The voltage sag of 0.6 pu is introduced at t= 0.1s to t= 0.25s whose magnitude is reduced to 138V as indicated in figure 1, at t= 0.1s to t= 0.25s.Its corresponding RMS voltage is shown in figure 2. The power system is connected to a nonlinear load of a three-phase bridge rectifier with R= 40Ω and L=1e-4 H through a transmission line of resistance 0.002 ohms and inductance of 1e-5 H. At t=0.02 s, the linear load is added to the system to analyze the effect of a sudden change in load on the power system considered. The decrease in voltage and increase in current from t= 0.02s to t=0.16s, due to the addition of load is shown in figure 3 and figure 4.

Fig.1. Voltage waveform without UPQC Compensation

Fig.2. RMS voltage (sag) without UPQC compensation due to a change in load

Fig.3. Voltage waveform to indicate a decrease in voltage due to a change in load

Fig.4. Current waveform to indicate an increase in current due to a change in load

The RMS current waveform for variation in load is as shown in figure 5 where the load is added at t=0.02 s to t= 0.16s. The load is removed at t=0.16s, and hence the magnitude of the current decreases at t=0.16s. At t= 0.1 s, the voltage sag of 0.6 pu is introduced which indicates the reduction in power at that instant as shown in figure 6. It is observed in figure 7, that the power factor without UPQC compensation is 0.94.

Fig.5. RMS Current waveform to indicate an increase in current due to a change in load

Fig.6. Waveform representing power before UPQC compensation

Fig.7. The waveform of Power factor without UPQC compensation

Design of unified power quality controller (UPQC)

In view of improving the system performance by mitigating the defined power quality issue in section 2, UPQC is designed with a unit vector algorithm for series active power filter (APF) and Instantaneous reactive power theory for shunt active power filter. The gate pulse is generated using the PWM technique in series APF and PWM and the hysteresis controller in the shunt active filter. The MATLAB simulation model designed for the defined power system network is shown in figure 8.

Three-phase voltage source in parallel to the microgrid at the supply terminal is connected to a three-phase load (both linear and nonlinear) through the transmission line. The UPQC comprising of series and shunt compensator is as shown in figure 8. The series active filter is designed using a unit vector control algorithm. The single phase input is fed to the phase-locked loop with the gain of 0.005 to obtain the value of wt. The three-phase voltage is derived by using the equation (1) – (3) which is then compared with the supply voltage to determine the error in voltage that derives the switching signals using the method of pulse width modulation.

Fig.8. A simulation model for Power system network with UPQC

.

The series active filter is connected to the supply terminal using three phases linear transformer with 12 terminals with three phases rated power of 10e6VA, with 50 Hz frequency. The winding RMS voltage is set to 100V, R= 0.002pu and X= 0.00005 pu.

The design of shunt APF is done using instantaneous reactive power theory (IRPT) with two different control algorithms for the generation of switching signals. One uses a PWM controller and the other uses a hysteresis controller. The IRPT is designed using supply voltage and load current. The reference current is obtained using supply voltage, load current, and control signal obtained using a PI controller. The reference current is obtained using the equation (4) – (9)

.

The series and shunt compensation is provided by using a MOSFET-based universal bridge inverter connected along with the interfacing inductor of 4e-3 H. The simulation results after providing UPQC compensation for the problem formulated are shown in figure 9. In figure 9, at t= 0.05s and t=0.15s, with a magnitude of voltage 231.6 V. Its corresponding RMS component of voltage is shown in figure 10.

Fig.9. Waveform of voltage after UPQC compensation

Fig.10. The waveform of RMS voltage after UPQC compensation

Fig.11. Waveform of current after UPQC compensation

The current waveform is as indicated in figure 11. The magnitude of current is 14.54 A at t= 0.02s to t=0.16s. At t=0.16s the linear load is removed and hence the magnitude of the current is reduced to 12.3A. After UPQC compensation the power waveform is as shown in figure 12. The magnitude of power is 1543 W at t= 0.02s to t= 0.16s. As the load is removed at t=0.16s, the power magnitude also reduces to 1333 W. The power factor after UPQC compensation is as shown in figure 13, where the power factor is improved to 0.98 compared to 0.94 without UPQC compensation. As per IEEE standards 519-1992: For general distribution system, current distortion limits obtained by the simulation performed lie within the limit.

Fig.12. The waveform of Power after UPQC compensation

Fig.13. Waveform of Power factor after UPQC compensation

The comparison of simulation output obtained from PWM and hysteresis controller for system without variation in load and with variation in load for different magnitudes of voltage sag and swell and harmonic orders are as tabulated in table 1 to table 6.

Table 1. THE MAGNITUDE OF FUNDAMENTAL COMPONENT OF VOLTAGE AND CORRESPONDING THD FOR VOLTAGE SAG AND SWELL WITHOUT LOAD CHANGE.

.

Table 2. THE MAGNITUDE OF FUNDAMENTAL COMPONENT OF CURRENT AND CORRESPONDING THD FOR VOLTAGE SAG AND SWELL WITHOUT LOAD CHANGE.

.

Table 3. THE MAGNITUDE OF FUNDAMENTAL VOLTAGE COMPONENT AND CORRESPONDING THD FOR VOLTAGE SAG AND SWELL WITH LOAD CHANGE.

.

Table 4. THE MAGNITUDE OF FUNDAMENTAL COMPONENT OF CURRENT AND CORRESPONDING THD FOR VOLTAGE SAG AND SWELL WITH LOAD CHANGE.

.

Table 5. THE MAGNITUDE OF FUNDAMENTAL COMPONENT OF CURRENT AND CORRESPONDING THD FOR HARMONICS

.

Table 6. THE MAGNITUDE OF FUNDAMENTAL COMPONENT OF VOLTAGE AND CORRESPONDING THD FOR HARMONICS.

.
Conclusion

The aim of power quality improvement is to provide pure sinusoidal voltage and current to the consumers. The problem formulated includes the addition of load at the distribution end, introducing voltage sag, swell, and harmonics of a different order which is as shown in figure 1 – figure 4.

The above-defined power quality-related issues are mitigated using a combination of shunt and series active power filters called UPQC with PWM and hysteresis controller as shown in figure 9-figure 11.The bar chart representation of THD for voltage and current without UPQC, with UPQC PWM and UPQC Hysteresis is shown in figure 14 –figure17. It can be concluded that PWM controller gives good performance when compared to hysteresis controller. The power and power factor, corresponding RMS voltage, and current is analyzed to show the effectiveness of UPQC designed to improve the power quality.

Fig.14. Voltage THD for power system without variation in load

Fig.15. Current THD for power system without variation in load

Fig.16. Voltage THD for power system with variation in load

Fig.17. Current THD for power system without variation in load

REFERENCES

[1] Priyank Shah, Ikhlaq Hussain, Bhim Singh, Ambrish Chandra, Kamal Al Haddad “GI Based Control Scheme for Single Stage Grid Interfaced SECS for Power Quality Improvement”, 2018, IEEE Transactions on Industry Applications, 1–1. DOI:10.1109/TIA.2018.2866375
[2] Sean Elphick, Vic Gosbell, Vic Smith, Sarath Perera, Philip Ciufo, Gerrard Drury, “Methods for Harmonic Analysis and Reporting in Future Grid Applications” 2016, IEEE Transactions on Power Delivery, 1–1. doi:10.1109/TPWRD.2016.2586963. Farnaz Harirchi, Marcelo Godoy Sim˜oes, “Enhanced Instantaneous Power Theory Decomposition for Power Quality Smart Converter Applications”, IEEE transactions on power electronics, vol.33, no.11, November 2018.
[3] Marco Mauri, Luisa Frosio and Gabriele Marchegiani,“ Electrical generation and distribution systems and power quality disturbance” , ISBN: 978-953-307-329-3, November 21st, 2011.Web of Science publication.
[4] Ali Keyhani, Mohammad N. Marwali, Min Dai, “Integration of green and renewable energy in electric power systems”, 19 November 2009, Print ISBN:9780470187760 |Online ISBN:9780470556771 DOI:10.1002/9780470556771, Copyright © 2010 John Wiley & Sons, Inc.
[5] Sachin Devassy, Bhim Singh, “Modified p-q Theory Based Control of Solar PV Integrated UPQC-S”, 2016 IEEE Transaction on Industry Applications, DOI: 10.1109/IAS33229.2016
[6] Alireza Javadi, Member, IEEE, Lyne Woodward, Member, IEEE, and Kamal Al-Haddad, Fellow, IEEE, “Real-time Implementation of a Three-phase THSeAF Based on VSC and P+R controller to Improve Power Quality of Weak Distribution Systems”, DOI 10.1109/TPEL.2017.2697821, IEEE transaction on Power Electronics. Pankaj D. Achlerkar; S. R. Samantaray; M. Sabarimalai Manikandan, “Variational Mode Decomposition and Decision Tree Based Detection and Classification of Power Quality Disturbances in Grid-Connected Distributed Generation System”, IEEE Transactions on Smart Grid ( Volume: 9, Issue: 4, July 2018), DOI: 10.1109/TSG.2016.2626469.
[7] Mingyu Lei, Student Member, IEEE, Zilong Yang, Yibo Wang, Honghua Xu, Lexuan Meng, Member, IEEE, Juan C. Vasquez, Senior Member, IEEE, Josep M. Guerrero, Fellow, IEEE, “An MPC Based ESS Control Method for PV Power Smoothing Applications”, IEEE transaction on Power Electronics, DOI:10.1109/TPEL.2017.2694448, PP 99,1-1, April 2017.
[8] Shafiuzzaman K. Khadema, Malabika Basu, Michael F. Conlon, “A comparative analysis of placement and control of UPQC, in DG integrated grid-connected network”, Sustainable Energy, Grids and Networks, Elsevier publication, 2016.
[9] Arya, Sabha Raj; Kant, Krishan; Niwas, Ram; Singh, Bhim; Chandra, Ambrish; Al-Haddad, Kamal (2014) “Power quality improvement in isolated distributed generating system using DSTATCOM”, IEEE 2014 IEEE Industry Applications Society Annual Meeting – Vancouver, BC, Canada (2014.10.5-2014.10.9)] 2014 IEEE Industry Application Society Annual Meeting -, (), 1–8. doi:10.1109/ias.2014.6978417
[10]Bhim Singh, Ambrish Chandra, and Kamal Al-Haddad, “Power Quality Problems and Mitigation Techniques”, First Edition. © 2015 John Wiley & Sons, Ltd. Published 2015 by John Wiley & Sons, Ltd.
[11]Vijayshree. G, “Improvement of Power Quality Using Instantaneous Reactive Power Theory for Controlling Shunt Active Filter,” 2021 International Conference on Circuits, Controls and Communications (CCUBE), 2021, pp. 1-6, doi: 10.1109/CCUBE53681.2021.9702742.

Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 99 NR 3/2023. doi:10.15199/48.2023.03.25

Upstream Frequency Disturbance

Published by Unipower AB, Metallgatan 4C, 44132 Alingsås, Sweden. Email: info@unipower.se

We have received questions about disturbances in the grid. Do we have any examples? Here is an interesting example showing how disturbances spread through the network.

A major frequency disturbance occured in Scandinavia on December 1, 2005. In this comparison we have compared measuring data from the source of disturbance in northern Sweden with data from our own meters in Alingsås (near Gothenburg) and we can also see the consequences in northern Norway.

The cause of the disturbance is disconnection of a large power plant in northern Sweden creating loss of critical power generation. This generates a major frequency dip. The time for the disturbance is 15:02:36 on December 1, 2005 and the frequency drops to 49.3 Hz. The disturbance is of the same magnitude in all of Sweden. The dip in Sweden was 0.7 Hz. In Norway, there is a frequency rush of 0.6 Hz (see the two top charts below). How can this be?

.

Well, at the time of the disturbance Norway is supplying power to Sweden. As a result from the disturbance, the connections between Norway and Sweden are disconnected. The effect is that Sweden, which is a large load, disappears and there is suddenly a great surplus of power in Norway. A large swell was registrered by the UP2210 as a result of the frequency disturbance. A swell of 8000 Volt is registrered more than 300 kilometres from the disturbance (see lower chart above).

The main consequences in Sweden were different types of breakdowns and production standstills. The process industry and heating plants were especially afflicted. Practically all heating plants had problems due to this disturbance.

About Unipower: Unipower AB offers a wide range of products for Power Quality measurements and Smart Grid systems.

Originating from a Swedish ABB company in the mid 80’s, Unipower has developed a competitive edge within the field of Power Quality and Smart-Grid solutions. We focus on norm compliance equipment, with a special focus on the requirements for power generation, transmission and distribution.

Our product lines reach from traditional portable PQ analysers to fully integrated and automated Power Quality Management systems for continuous supervision of the energy supply.

Website: unipower.se

Source URL: https://www.unipower.se/about-power-quality/upstream-frequency-disturbance/

Evaluating Direction of Harmonics – Power Harmonics

Published by Unipower AB, Metallgatan 4C, 44132 Alingsås, Sweden. Email: info@unipower.se

One important issue is how to trace the source of harmonics. The example below shows harmonic filters connected until 11.00 and there is no voltage distortion as expected. After 11.00 the filters were disconnected due to a fault and voltage harmonics can be seen.

There are at least 2 ways to determine the direction of harmonics

.

First is to simply look at current with current harmonics and compare with voltage. If there is obvious correlation between current and voltage harmonics, then the source is downstream i.e. the load current is causing the voltage distortion.

The second, and easier, way to study harmonics direction is to study the power harmonics. For each frequency voltage and current creates active power, P. With normal references the sign of P tells the direction of power flow, + means from grid to load (source upstream) and – means from load to grid (source downstream).

Below are the power harmonics from the above example.

.

You clearly see that 5th (green) and 7th (red) harmonics have negative sign meaning a downstream direction, i.e. going from load into the grid.

About Unipower: Unipower AB offers a wide range of products for Power Quality measurements and Smart Grid systems.

Originating from a Swedish ABB company in the mid 80’s, Unipower has developed a competitive edge within the field of Power Quality and Smart-Grid solutions. We focus on norm compliance equipment, with a special focus on the requirements for power generation, transmission and distribution.

Our product lines reach from traditional portable PQ analysers to fully integrated and automated Power Quality Management systems for continuous supervision of the energy supply.

Website: unipower.se

Source URL: https://www.unipower.se/about-power-quality/evaluating-direction-of-harmonics/

Impact of Ambient Temperature on the Failure Intensity of Overhead MV Power Lines

Published by Andrzej Ł. Chojnacki, Kielce University of Technology, Department of Power Engineering
ORCID: 0000-0002-9227-7538

Abstract. The article presents the influence of weather conditions, represented by the ambient temperature, for the intensity of the failure of MV overhead power lines. It presents the mechanisms of damaging the equipment by the high and low temperatures. It shows the method of determining the intensity of failure of power facilities as a function of ambient temperature. The article presents the empirical results obtained for the MV overhead power lines exploited in the Polish electricity grids.

Streszczenie. W artykule przedstawiono wpływ warunków atmosferycznych reprezentowanych przez temperaturę otoczenia na intensywność uszkodzeń napowietrznych linii elektroenergetycznych SN. Zaprezentowano w nim mechanizm uszkadzania tych urządzeń na skutek oddziaływania wysokich oraz niskich temperatur. Omówiono metodę modelowania zależności intensywności awarii obiektów energetycznych od temperatury otoczenia. Zaprezentowano wyniki uzyskane podczas wieloletnich badań dla napowietrznych linii elektroenergetycznych SN eksploatowanych w polskich sieciach dystrybucyjnych energii elektrycznej. (Wpływ temperatury otoczenia na intensywność awarii napowietrznych linii elektroenergetycznych średniego napięcia)

Słowa kluczowe: temperatura otoczenia, intensywność awarii, linie dystrybucyjne, niezawodność systemu elektroenergetycznego
Keywords: ambient temperature, failure intensity, power distribution lines, power system reliability

Introduction

A contemporary recipient of electricity sets very high requirements concerning the quality and continuity of electricity supply. The systematically increasing unit power rating of power stations and lines increases the danger of shutdown of higher power values in the case of their failure, thus leading to increasing limitations in electricity supply for the recipients. This causes substantial material losses and in extreme cases can result in health or life hazards. In order to avoid the aforementioned hazards, design engineers must know the principles applicable to power device reliability and aim at optimally selecting the device parameters, thereby ensuring reliable operation. The timeliness of the aforementioned aspects and the need to conduct further research in the scope of the reliability of power structures are confirmed, among others, by numerous publications on the subject matter [9, 13, 16].

According to the definition presented in [12, 14], reliability is the ability of elements (structures) to perform the set functions in specific conditions and in a determined time period, with simultaneous adherence to acceptable parameters. Usually, “specific conditions” are adopted as fixed, and reliability is considered only in the operation time function. Meanwhile, it is necessary to note that time does not directly affect the reliability of structures. Any changes in the ability of elements to perform the set functions are an effect of internal and external (environmental) exposure. Exposure changes in time and the changes are usually random.

The impact of the environment on the behaviour of structures has been known for a long time. Many years before the development of reliability theory, standardisation acts on environmental studies were established, aimed at checking whether the structure is able to perform its task if the specified environmental exposure will affect the structure with specific intensity and for a specific time period. When designing structure reliability, two components are often taken into consideration: temperature impact and total impact of other environmental exposure [10, 12]. Unfortunately, the literature features a relatively low number of up-to-date elaborations on the topic. It is more often possible to find publications concerning the impact of weather on the variation of electrical loads, e.g. [1].

Impact of ambient temperature on the reliability of power engineering structures

When analysing the impact of ambient temperature on the reliability of power engineering structures, it is necessary to take into consideration three principal aspects

• impact of high temperature;
• impact of low temperature;
• impact of quick changes in temperature

The maximum air temperature in the shade, in open areas, does not exceed 60o C on the ground, but the surface temperature of devices located in open areas, without covers, can exceed 100oC. Such temperatures have a negative impact on the operation of particular structures and their components. High temperature can be the cause of considerable damage, because it degrades material properties by causing their softening, melting, sublimation, evaporation, reduction in viscosity, changes in sizes and thermal ageing [3, 10, 15, 18]. High temperature can cause mechanical deformations resulting from the expandability of materials. They are especially strong in the case of combining materials with various expandability factors or uneven heating of parts of a single material, but with substantial sizes. Mechanical deformation can in turn be the cause of mechanical damage or changes in products’ electrical parameters [10]. On the other hand, the softening and melting of plastic materials leads to structural weakening or damage, and to resin fill leaks. Accelerated thermal ageing of organic insulation materials manifests itself with the migration of softeners and their evaporation from thermoplastic materials. On the other hand, thermosetting materials lose their volatiles and become delaminated. Material ageing can also cause changes in their electrical properties, namely [4, 10, 15]:

• reduction in the electrical cross-resistivity and surface resistivity of dielectrics;
• reduction in the voltage resilience of dielectrics;
• changes in the dielectric constant of dielectrics (the character of changes and their magnitude vary for different materials);
• increase in dielectric loss;
• increase in the electrical resistivity of metals.

The structural resistance to high temperature impact can be verified in laboratory conditions by conducting a dry heat test according to the IEC 60068-2-2:2007 standard [7].

Negative air temperatures can also have a negative impact on power devices, especially because the air temperature of devices placed in the open air can reach values significantly below the ambient temperature as a result of heat radiation. Negative temperatures mainly contribute to the degradation of the mechanical properties of power devices and structures. They cause an increase in the fragility of materials, increase in viscosity and solidification of liquids, reduction in mechanical durability and shrinking of materials. Changes in the linear size cause mechanical damage by the jamming and seizure of co-operating movable components. Material shrinkage can also contribute to the weakening of joints and part breakage or cracking. As result of frosting and icing, the product’s weight increases, which can also cause damage. Plastics usually become hardened and more fragile in negative temperatures. Changes in the dimensions and hardness of washers and gaskets can cause the unsealing of products. Solders with high tin content become fragmented and crushed under weak impact. Increases in the viscosity of lubricants and oils hinder the operation of movable components, and their damage in the case of lubricant freezing [10].

Negative temperatures also lead to changes in the electrical parameters of materials, such as electrical conductivity, dielectric loss, dielectric constant and magnetic permeability, thereby causing changes in the parameters of electrical elements and devices, whereas these are usually beneficial changes (reduction in the dielectric loss factor of many insulating materials, increase in insulation’s electrical durability and resistivity, along with a reduction in temperature) [10].

The structural resistance to negative temperature can be verified in laboratory conditions by conducting a cold test according to the IEC 60068-2-1:2007 standard [5].

Changes in temperature result from daily changes in air temperature, varying insolation, sudden wetting of the device, etc. Daily temperature changes are substantially smaller at the seaside than deep inland. Sudden temperature changes affect power devices and structures subjected to the direct impact of solar radiation. Their surface can reach temperatures exceeding 100oC, after which they become wetted by rain with a substantially lower temperature (the temperature of hailstorms can reach approx. 0oC). In this case, the surface temperature can change rapidly by approx. 100oC [10, 15].

Changes in temperature can cause dangerous mechanical stress in the material and cause changes in electrical parameters. Fast expansion and shrinking of materials leads to the weakening of connections and to cracks and breakage. In the case of sealed devices, they can become unsealed. A common phenomenon is the cracking of protective covers as a result of the variable expandability of the base and protective layer’s material. The effects of fast temperature changes are similar to those caused by positive and negative temperatures. The difference is that high intensity of temperature changes causes more complete damage, whereas positive or negative temperatures cause parameter damage (partial, incomplete) far more often [10, 15].

Structural resistance to the impact of temperature changes can be verified in laboratory conditions by conducting the tests specified in the IEC 60068-2-14:2009 [6] and IEC 60068-2-33:2002 [8] standards.

Seasonality and causes of damage to overhead mv power lines

Table 1 presents the frequency of damage to overhead MV power lines with distribution to particular months. Data in the form of histograms and approximate functions are presented in figure 1.

Table 1. Frequency of overhead MV line failures in particular months of the year [%]

.
Fig.1. Empirical values and approximate function of the seasonal variance in the frequency of failures of overhead MV lines

Most failures of overhead MV lines were observed in the summer (July, August) and in winter (December, January). The summer featured 427 failures, which constitute 21.90% of all damage. The winter featured 406 failures, which constitute 20.82% of all damage. In other months, the lines’ unreliability is below the average damage intensity. This allows for the general statement that extreme temperatures occurring in the winter and summer periods greatly affect the failure intensity in power devices.

Any mathematical functions can be the approximate function. Due to the transparency and simplicity of the transcript, a multinomial was adopted as the approximate function. Due to the fact that the approximate function factors obtained for an order higher than fourth are close to zero, a decision was made to approximate the function of seasonal variability of line failure frequency with a multinomial of at least the fourth-order. The multinomial has the following form:

.

where: k – subsequent month number; a, b, c, d, e – approximate function factor.

The approximate function factors of the seasonal variability in the overhead MV line failure frequency presented in figure 1 amount to: a = 0.0137; b = -0.3832; c = 3.6405; d = -13.1109; e = 21.5470. The correlation factor of the designated function in relation to empirical data amounts to r = 0.85.

The percentage share of the causes of line failure with consideration of seasonality is included in table 2. The percentage share of particular causes of line failure in the total number of failures is presented in figure 2.

The most serious cause for overhead MV line failures is the ageing process, which contributes to approx. 19.38% of all line damage. The second cause concerns trees and branches, which resulted in approx. 16.31% of all damage. Seasonal causes that greatly contribute to the failure rate of overhead MV lines are electrical discharges as well as icing and frosting. They caused, respectively, 13.64% and 9.23% of all damage.

Table 2. Causes of overhead MV line failures in particular months [%]

.

Table 3. Length of overhead MV power lines in subsequent observation years [km]

.
Analysis of the impact of ambient temperature on the failure intensity in overhead mv power lines

The average intensity of power line damage, with the assumption of immobility and ergodicity of the damage and restoration processes, can be determined with the following dependency [2, 11, 17]:

.

where: m – observed number of failures in a time interval Δt; np – sample size at the beginning of the observation period; nk – sample size at the end of the observation period; Δt – total observation time.

Fig.2. Percentage share of the causes of overhead MV line failures

In order to designate the intensity’s temperature characteristics ¯λ = f(t) it is necessary designate the value ¯λ(Ti) for subsequent temperature ranges Ti. For this purpose, the dependency (2) takes into consideration the number of failures m(Ti) that occurred in a specific temperature range i and the duration of the temperature range Δt(Ti):

.

where: m(Ti) – number of failures in the time interval Δt(Ti); Δt(Ti) – time of occurrence of temperature Ti in the considered time period Δt; 𝜏(Ti) – relative time of occurrence of temperature Ti in the considered time period Δt:

.

When designating the value ¯λ(Ti) for subsequent ranges i, the empirical dependency of failure intensity on the ambient temperature is obtained.

Table 4. Results of calculations of failure intensity in overhead MV power lines depending on ambient temperature

.

The analysis of the impact of ambient temperature on the failure intensity in overhead MV lines was conducted based on the data derived from 15 years of observation conducted within the premises of a large power distribution company in Poland. At the beginning of the observation a total of 1,050 km of overhead MV lines were used in the company. At the end of the observation the length amounted to 1,211 km.

The length of the lines in particular observation lines is presented in table 3. During the 15 years of observation a total of 1,950 failures in overhead MV lines occurred.

Table 4 presents the results of the conducted analysis of the dependency of failure intensity in overhead MV lines on ambient temperature.

The failure intensity in overhead MV power lines depending on ambient temperature is presented in figure 3.

Determining the empirical waveform of function ¯λ(Ti) does not exhaust the problem of studying the dependency of failure intensity on ambient temperature. It is also important to determine the functional form of this dependency. The approximate function of the damage frequency presented in figure 3 is a fourth-degree multinomial expressed in the dependency (1), whereas k means the ambient temperature T. The approximate function factors of the line failure intensity in the ambient temperature function amount to: a = 21.29·10-6; b = -522.76·10-6; c = 18,110.54·10-6; d = 3,406.03·10-6; e = 8.8767. The correlation factor of the theoretical function with empirical data amounts to r = 0.91.

Fig.3. Dependency of the failure intensity of overhead MV power lines on ambient temperature

The hypothesis on the functional form was subjected to verification based on the character test and series test. As a result of using the character test, the following was obtained: l0 = min(l+ ,l– ) = min(15,12) = 12; l0 = 12 > 7 = la; lo ∊ Ra = (-∞,7). Therefore, at the weight level a = 0.05, there is no basis for rejecting the hypothesis on the functional form of distribution ¯λ = f(t). As result of using the series test, the following was obtained: n1 = l+ = 15; n2 = l– = 12; number of subsequent series k = 11. For the above values, the critical area covers the values k ≤ 9 and k ≥ 18. Therefore, at the weight level a = 0.05, there is no basis for rejecting the hypothesis on the functional form of distribution ¯λ = f(t).

Summary and final conclusions

The dependency of the failure intensity of power lines on the ambient temperature must be related to many factors. First of all, there is a considerable correlation between high temperatures and atmospheric discharges, and between low temperatures and icing and frosting on devices. These are factors that cause the vast majority of failures. Furthermore, high temperatures cause more difficult conditions of cooling the devices, which in the case of substantial loads leads to exceeding the acceptable temperatures. On the other hand, low temperatures lower the plasticity of most insulation and conductor materials. This causes that even small external forces can damage them. Lubricant materials used in movable connections also lose their properties. In connection with the reduction in the dimensions of elements constituting a movable connection (temperature expandability), this causes substantial resistance during movement and can lead to damage. A very dangerous phenomenon occurring at high temperatures, especially for porcelain elements (insulators, lightning rods), is their rapid cooling by rain after previous long-term heating. An analogous situation takes place for lightning rods operating at very low temperatures. In the case of arc ignition (e.g. due to a grid surge) during rapid cooling of the lighting rod’s insulator, it can be damaged due to substantial differences in the internal and external surface temperatures. Other phenomena that negatively affect the technical condition of devices and are related to extreme ambient temperatures are periodic temperature changes. In the hot summer period device elements become heated during the day. At night the temperatures are substantially lower, thus causing their cooling. Such periodic changes, with uneven temperature expandability of various materials, can lead to unsealing (e.g. transformer ladle) or loosening (e.g. connector contacts) of power device subassemblies. In the case of laminated insulation systems (e.g. cable heads, couplers, etc.) a possible damage mechanism at low temperatures is the destruction of insulation as a result of freezing of the moisture present in it. On the other hand, in periods of low temperatures the length of overhead line cables is reduced, thereby increasing their tension. Failures can occur in the case of any defects of the support structures or insulators. Low temperatures often cause problems with compressed air systems, especially with pressure valves that freeze up and are damaged if activated. An analogous situation takes place in the case of movable components of power devices. At low temperatures, they freeze to one another and become damaged at the attempt of mutual movement.

The impact of ambient temperature on the failure rate of power devices has been noticed for many years. However, no research aimed at determining this dependency in qualitative terms was conducted. The paper presents a method of designating the failure intensity of power structures in the ambient temperature function. The method was used to determine the dependency of failure intensity of overhead MV lines on the temperature at which they are used. The test results presented in the paper do not exhaust the impact of ambient conditions on the reliability of power structures. In further research, the author will attempt to develop multi-dimensional reliability models covering, aside from temperature, factors such as humidity, pressure, wind velocity, precipitation, atmospheric discharges, etc.

LITERATURA

[1] Bolzern P., Fronza G., Role of weather inputs in shortterm forecasting of electric load. International Journal of Electrical Power & Energy Systems, Volume 8, Issue 1, January 1986, Pages 42-46
[2] Chojnacki A. Ł., Analysis of the operating reliability of power distribution grids. Kielce University of Technology Publisher, Kielce, 2013
[3] Collective Work. Electrical Insulation Materials. Scientific and technical publishing house, Warszawa, 1965
[4] Collective Work. Electronics and telecommunication problems: Climate resilience and mechanical durability of electronic equipment. Wydawnictwo Komunikacji i Łączności, Warszawa 1968
[5] IEC 60068-2-1:2007 Environmental testing – Part 2-1: Tests – Test A: Cold.
[6] IEC 60068-2-14:2009 Environmental testing – Part 2-14: Tests – Test N: Change of temperature.
[7] IEC 60068-2-2:2007 Environmental testing – Part 2-2: Tests – Test B: Dry heat.
[8] IEC 60068-2-33:2002 Environmental testing – Part 2: Tests. Guidance on change of temperature tests.
[9] Johnson M., Gorospe G., Landry J., Schuster A., Review of mitigation technologies for terrestrial power grids against space weather effects. International Journal of Electrical Power & Energy Systems, Volume 82, November 2016, Pages 382-391
[10] Migdalski J. red., Reliability Engineering – Guide. ATR Bydgoszcz i Zetom Warszawa, 1992
[11] Migdalski J. red., Reliability guide. Mathematical basics. Wydawnictwo „WEMA”, Warszawa, 1982
[12] Military Standardization Handbook. Reliability Prediction of Electronic Equipment. MIL-HDBK 217B. U.S. Government Printing Office, Washington, 1974
[13] Narimani A., Nourbakhsh G., Ledwich G. F., Walker G. R., Optimum electricity purchase scheduling for aggregator storage in a reliability framework for rural distribution networks. International Journal of Electrical Power & Energy Systems, Volume 94, January 2018, Pages 363-373
[14] PN-N-50191:1993 Terminology of electrics – Reliability, quality of service.
[15] Rychtera M., Bartakova B., Tropic-proofing of electrical devices. Scientific and technical publishing house, Warszawa, 1966
[16] Sousa B. J. O., Humayun M., Pihkala A., Lehtonen M. I., Three-layer seasonal reliability analysis in meshed overhead and underground subtransmission networks in the presence of co-generation. International Journal of Electrical Power & Energy Systems, Volume 63, December 2014, Pages 555-564
[17] Sozański J., Reliability of electricity power supply. Scientific and technical publishing house, Warszawa, 1982
[18] Wróblewski Z., Multi-variant method of forecasting the durability of electro-magnetic AC contacts in current production. Wrocław University of Technology Publishing Houses, Wrocław, 1988

Autor: dr hab. inż. Andrzej Ł. Chojnacki, prof. PŚk, Politechnika Świętokrzyska w Kielcach, Katedra Energetyki, Energoelektroniki i Maszyn Elektrycznych, Aleja Tysiąclecia Państwa Polskiego 7, 25-314 Kielce, e-mail: a.chojnacki@tu.kielce.pl

Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 98 NR 10/2022. doi:10.15199/48.2022.10.68

Active and Reactive Power Control in a Three-Phase Photovoltaic Inverter

Published by 1.Adnan Majeed Abed1, 2. Afaneen Anwer Alkhazraji2, 3. Shatha S. Abdulla3, University of Technology, Iraq, Baghdad (1), University of Technology, Iraq, Baghdad (2), University of Technology, Iraq, Baghdad (3) ORCID: 1.0000-0002-6558-9552; 2.0000-0003-3995-8307;3. 0000-0001-5493-2485

Abstract. In most nations, grid-connected buildings with solar systems are expanding. Several sites in the system network have high PV penetration. The irregular nature of PV installations could affect the distribution network. Instead of expensive grid installations, PV systems can employ a voltage source inverter to utilize reactive power. The major objective is to inject and control 100 kW of three-phase, two-stage solar PV power into the grid in order to maintain a constant voltage independent of variations in solar radiation and to keep the current’s THD within international standards. PV system implementation depends on practical system concerns. Reactive power control and inverter control are created. The network variable the whole system shows good usage of reactive power. The suggested 100 KW PV system in this study achieves reactive power regulation and sinusoidal three-phase output currents. Using MATLAB 2021b and Simulink software, the recommended system’s effectiveness was elucidated and its viability was demonstrated. The results demonstrated the effectiveness of the recommended design and modelling.

Streszczenie. W większości krajów rozbudowuje się budynki podłączone do sieci z systemami słonecznymi. Kilka lokalizacji w sieci systemowej ma wysoką penetrację PV. Nieregularny charakter instalacji fotowoltaicznych może mieć wpływ na sieć dystrybucyjną. Zamiast drogich instalacji sieciowych, systemy fotowoltaiczne mogą wykorzystywać falownik źródła napięcia do wykorzystania mocy biernej. Głównym celem jest wprowadzenie i sterowanie 100 kW trójfazowej, dwustopniowej energii fotowoltaicznej do sieci w celu utrzymania stałego napięcia niezależnego od zmian promieniowania słonecznego oraz utrzymania THD prądu zgodnie z międzynarodowymi standardami. Wdrożenie systemu fotowoltaicznego zależy od praktycznych problemów systemowych. Tworzone jest sterowanie mocą bierną i sterowanie falownikiem. Zmienna sieciowa całego systemu wskazuje na dobre wykorzystanie mocy biernej. Sugerowany w tym badaniu system fotowoltaiczny o mocy 100 KW zapewnia regulację mocy biernej i sinusoidalne trójfazowe prądy wyjściowe. Za pomocą oprogramowania MATLAB 2021b i Simulink wyjaśniono skuteczność zalecanego systemu i wykazano jego żywotność. Wyniki wykazały skuteczność zalecanego projektu i modelowania. (Sterowanie mocą czynną i bierną w trójfazowym falowniku fotowoltaicznym)

Keywords: Grid-connected photovoltaic inverters; maximum, power point tracking; power and reactive power control.
Słowa kluczowe: Inwertery fotowoltaiczne, podłączone do sieci; maksymalny, śledzenie punktów mocy; sterowanie mocą i mocą bierną.

1. Introduction

Renewable energy sources for home use are becoming increasingly popular because of their quiet operation and environmental friendliness [1]. Since The most efficient way to use solar-generated electricity is to feed it directly into the air conditioner, it is impossible to have a PV power system without an inverter that is connected to the grid [2]. Gadget number two, a PV inverter, may also be a viable option [3]. Reactive power is required to increase the electrical grid’s capacity. Consequently, a PV inverter providing reactive power is necessary. A PV power system that is currently in use needs a dependable power source to function [4]. The most powerful system is the PV power conditioning unit. The maximum power point tracking (MPPT) control mechanism is crucial for harvesting power at the maximum amount that the solar array may provide. [5]. The current regulator must be as efficient as feasible in order to work. It has to be in place in order to maintain proper control over both reactive and active power [6]. The phase-locked loop (PLL) is also necessary for system-wide voltage and current control, including the regulation of phase and amplitude according to the grid frequency. Furthermore, the total array’s maximum power output should be plugged into the electricity grid.

In this research, it is provided the q-axis and d-axis current commands for the targeted reactive power with a maximum output PV array. Feedforward and feedback controllers can be used to manage these currents. The Maximum Probability of Success (MPPT) algorithm [5]. In this experiment, perturbation and observation techniques are employed [7]. Being capable of the space vector, as well as data about grid voltage and current. A phase-locked loop is also shown. Guidelines for using the PLL and current controller are shown. Additionally, a design concept, an analysis, and a simulation are all necessary components detailed. Finally, the results of the simulations support the proposal for control mechanisms.

The subsequent portions of the study are organized as follows: The second section outlines a proposed system. There is discussion of the MPPT algorithm, power control, PLL design, and current controller design. Sections 3 and 4 analyse and summarize the simulation’s results.

2. System of Photovoltaic Power Conditioning

Figure 1 depicts the circuit architecture for the three-phase grid-connected PV inverters. The PV array, boost converter, DC connection, and inverter make up the inverter. The MPPT controls the boost converter. The transfer of control of the grid’s active and reactive functions is powered by a three-phase inverter.

Fig.1. The grid-connected, three-phase PV inverters’ electrical circuitry.

The boost converter and switching frequency of the three-phase inverter are defined for the 380V/50Hz three-phase PV power conditioning system.

2.1 MPPT Algorithm

In this section, it will be discussed the perturbation and observation-based MPPT approaches in depth [4]. With the help of these methods, it can determine the current power output of solar panels. An increase in duty cycle may be appropriate in some locations but not others (D). For optimal performance, the converter must be raised or lowered such that it operates near its maximum power point. Figure 2 shows the outcome of a straightforward PV power comparison, or straightforward MPPT, in which Ppv (k) denotes the PV power that is now in use and Ppv (k-1) denotes the previous number.

2.1 MPPT Algorithm In this section, it will be discussed the perturbation and observation-based MPPT approaches in depth [4]. With the help of these methods, it can determine the current power output of solar panels. An increase in duty cycle may be appropriate in some locations but not others (D). For optimal performance, the converter must be raised or lowered such that it operates near its maximum power point. Figure 2 shows the outcome of a straightforward PV power comparison, or straightforward MPPT, in which Ppv (k) denotes the PV power that is now in use and Ppv (k-1) denotes the previous number.

Fig.2. The MPPT method.

Fig.3. P & O MPPT flowchart.

2.2 Power Control: Active and Reactive

In this subsection, the suggested active and reactive power regulation is introduced. The transformation matrices between the phase and space vectors in Fig. 4 are 3/2 and 2/3 according to (1) and (2), respectively. where f denotes the current or voltage [8].

.
.

Figure 4 displays the equivalent representations of the rotating and stationary frames, matrices for the axis transformation and its inverse. Where “→” denotes the spatial vector component, d and q denote the rotating reference frame’s components, and α, β denote the fixed reference frame’s components. Grid voltage has an angular frequency of ω (= dθ / dt) [9].

Fig.4. The suggested control makes use of two coordinate systems.

.

The stationary reference frame’s active power (P) and reactive power (Q) are calculated using (5). When the rotating reference frame and grid voltage are synced, the rotating reference frame’s vq equals zero, and P and Q may then be determined using (6) [10].

.

As noted, the d-axis current or active power order (𝑖*d) is recommended to ensure the grid obtains the maximum amount of electricity possible (7). The q-axis current or reactive power order (𝑖*q) is gathered in (6) in order to manage reactive power by regulating the reactive power (8) [11].

.

According to (9) and (10), respectively, Figure 4 shows the state equations for both stationary and rotating reference frames. Where “*” signifies the ordered value and “^” indicates the PLL’s approximate amount [12].

.

The voltage orders are shown in equation (11), which is what equation (10) becomes in a stable state (12). Where v′d and v′q are the current controller voltages on the d- and q-axes, respectively.

.

where: 𝑉’d– feedback term, 𝑉gd – ŵLiq– feedforward term, 𝑉’q – feedback term, ŵLid– feedforward term.

2.3 Space vector PLL design

Since PLL management is required by the utility system to synchronize inverter voltage output to the connected utilities, this section provides an overview of the space vector PLL. Grid phase voltages are located and converted into a space vector quantity. when the grid voltage and the rotating frame are synchronized, the vq on the rotating frame equals zero. As a result, the PI controller controls the vq to zero. The space vector PLL block diagram is seen in Fig. 5. The vq’s small-signal is examined to be able to construct the PI controller. When the PLL is in control of vq, a relevance of vq is displayed in (13).

.

Because the Δθ is quite small near the balance operating conditions, the sin (Δθ) is about Δθ. A space vector PLL’s small-signal circuit diagram is shown in Fig. 6.

Fig.5. The space vector PLL block diagram.

Fig.6. Space vector PLL’s small-signal block diagram

2.4 Current Controller Design This subsection discusses current control as a way to regulate the ordered active and reactive power [13]. Figure 7 depicts the voltage command that drives the d-axis current loop (12). Noting that the PLL control and loop-transfer functions are similar, the PI controller was designed in a similar fashion. The crossover frequency (w0) is determined to be 1000 rad/s in order to demand the risetime (tr) of approximately 2.2 ms using the equation tr ≈ 2.2 / w0 and a corner frequency equal to 25% of the crossover frequency is chosen. Then, the kp and ki are, respectively, 10 and 50000. The identical value uses the q-axis current controller’s PI gain.

Fig.7. the current loop on the d-axis.

Fig.8. The suggested control strategies.

The proposed control strategies are shown in Fig. 8.

3. Results of the simulation

In this part, it will be described the simulation procedure used to evaluate the performance of suggested control schemes. The MATLAB R2021b simulation program is used.

Table 1. System design requirements

.

At maximum power (100 KW) and average solar intensity (1000 W/m2), the photovoltaic modules’ voltage and current are 290V and 345.45A, respectively. In Figs. 9 and 10, the simulation values are presented.

Investigate 1: The fundamental waveforms of the proposed PV inverter are displayed in Fig. 9 for a variety of reactive powers and a constant active power. Modifications are made so that there is 1000 W/m2 of solar irradiation (S).

Figure 9 shows the voltage, current, and reactive power (Q) injected or absorbed for the three-phase grid. The Q* has a value of zero at the beginning. At time interval t1, the Q* abruptly increases from 0 to 10 kvar. Therefore, as the reactive power command rises, the igu’s magnitude and phase angle do as well. The Q suddenly takes 10 kvar from the grid at time interval t3. At times t2 and t4, 100 KW of active electricity are sent into the grid by the grid-connected inverter of the PV system, which operates at a unity power factor. It is discovered that the suggested control methods can smoothly manage the reactive output power of the PV inverter without severely reducing active power.

Investigate 2: In Fig. 10, the primary waveforms of the suggested PV inverter are shown when it is operating with a constant reactive power of zero and under varied active power or solar intensity circumstances (S).

Fig.9. The suggested PV inverter’s main waveforms feature constant active power but varying reactive power.

Figure 10 displays the three-phase grid’s voltage, current, and active power output. At time interval t5, the solar irradiance (S) was originally set at 1000 W/m2, or 100 KW of active power command. S rapidly falls from 1000 W/m2 to 500 W/m2. The active power command’s step causes the Ppv and P to rapidly change to 500 KW, as planned, and the amplitude of grid currents to similarly abruptly change, without oscillation. It should be observed that the MPPT operation transitions smoothly and reacts quickly to the peak power point. The harmonic spectra of the u-phase inverter at steady state operation with S = 1000 W/m2 is depicted in Fig. 11. The u-phase inverter has a very low THDi of 1.41 percent.

Fig.10. The proposed PV inverter’s primary waveforms have varying active powers but a constant reactive power of zero.

Fig.11. the grid current’s harmonic spectrum in u-phase.

4. Conclusion

An easier three-phase grid-connected PV inverter with reliable active and reactive power management, minimal current harmonics, seamless transitions, and quick response to MPPT control’s maximum power point was described in this study. Results from simulations were utilized to support this. The simulations also demonstrate the value of PLL and the most recent standards for controller design.

REFERENCES

[1] Afaneen A. Abbood, Mohammed A. Salih and Hassan N. Muslim “Management of Electricity Peak Load for Residential Sector in Baghdad City by Using Solar Generation” International Journal of Energy and Environment, Vol 8, Issue 1, pp.63-72, 2017.
[2] H. J. Avelar, W. A. Parreira, J. B. Vieira, L. C. G. De Freitas, and E. A. A. Coelho, “A state equation model of a single-phase grid-connected inverter using a droop control scheme with extra phase shift control action,” IEEE Trans. Ind. Electron., vol. 59, no. 3, pp. 1527–1537, 2012, doi: 10.1109/TIE.2011.2163372.
[3] A. A. Abbood, M. A. Salih, and A. Y. Mohammed, “INTERNATIONAL JOURNAL OF ENERGY AND ENVIRONMENT Modeling and simulation of 1mw grid connected photovoltaic system in Karbala city,” J. homepage http://www.IJEE.IEEFoundation.org ISSN, vol. 9, no. 2, pp. 2076– 2909, 2018, [Online]. Available: http://www.IJEE.IEEFoundation.org.
[4] A. Cagnano, E. De Tuglie, M. Liserre, and R. A. Mastromauro, “Online optimal reactive power control strategy of PV inverters,” IEEE Trans. Ind. Electron., vol. 58, no. 10, pp. 4549–4558, 2011, doi: 10.1109/TIE.2011.2116757.
[5] T. Kok Soon, S. Mekhilef, and A. Safari, “Simple and low-cost incremental conductance maximum power point tracking using buck-boost converter,” J. Renew. Sustain. Energy, vol. 5, no. 2, 2013, doi: 10.1063/1.4794749.
[6] T. Aung and T. L. Naing, “DC-link voltage control of DC-DC boost converter-inverter system with PI controller,” World Acad. Sci. Eng. Technol. Int. J. Electr. Comput. Eng., vol. 12, no. 11, pp. 848–856, 2018.
[7] R. I. Putri, S. Wibowo, and M. Rifa’i, “Maximum power point tracking for photovoltaic using incremental conductance method,” Energy Procedia, vol. 68, pp. 22–30, 2015, doi: 10.1016/j.egypro.2015.03.228.
[8] F. Blaabjerg, R. Teodorescu, M. Liserre, and A. V. Timbus, “Overview of control and grid synchronization for distributed power generation systems,” IEEE Trans. Ind. Electron., vol. 53, no. 5, pp. 1398–1409, 2006, doi: 10.1109/TIE.2006.881997.
[9] O. Rabiaa, B. H. Mouna, S. Lassaad, F. Aymen, and A. Aicha, “Cascade Control Loop of DC-DC Boost Converter Using PI Controller,” Int. Symp. Adv. Electr. Commun. Technol. ISAECT 2018 – Proc., no. Ccm, pp. 1–5, 2019, doi: 10.1109/ISAECT.2018.8618859.
[10] T. I. Suyata, S. Po-Ngam, and C. Tarasantisuk, “The active power and reactive power control for three-phase gridconnected photovoltaic inverters,” ECTI-CON 2015 – 2015 12th Int. Conf. Electr. Eng. Comput. Telecommun. Inf. Technol., vol.2, no. 1, 2015, doi: 10.1109/ECTICon.2015.7207066.
[11] T. Huang, X. Shi, Y. Sun, and D. Wang, “Three-phase photovoltaic grid-connected inverter based on feedforward decoupling control,” ICMREE 2013 – Proc. 2013 Int. Conf. Mater. Renew. Energy Environ., vol. 2, pp. 476–480, 2013, doi: 10.1109/ICMREE.2013.6893714.
[12] R. Benadli, B. Khiari, and A. Sellami, “Three-phase gridconnected photovoltaic system with maximum power point tracking technique based on voltage-oriented control and using sliding mode controller,” 2015 6th Int. Renew. Energy Congr. IREC 2015, pp. 0–5, 2015, doi: 10.1109/IREC.2015.7110963.
[13] H. A. Hussein, A. J. Mahdi, and T. M. Abdul-Wahhab, “CurrentControl Inverter Schemes for a Grid – Connected PV Generator,” IOP Conf. Ser. Mater. Sci. Eng., vol. 1105, no. 1, p. 012018, 2021, doi: 10.1088/1757-899x/1105/1/012018.

Authors: Adnan Majeed Abed, received the Engineer degree in electrical Engineering from University of Technology, BaghdadIraq, in 2008. He also has a BA in English from the College of Languages, University of Baghdad, Iraq in 2013. He has been working as an electrical engineer in the Iraqi Ministry of Electricity
since 2008 until now. You can connect with me by this e-mail: eee.20.55@grad.uotechnology.edu.iq .
Afaneen Anwer Alkhazraji was born in Baghdad, lraq in January of 1967.she received her B. Sc in 1990 from Baghdad university. Her M. Sc and PhD degrees in 1998 and 2005 respectively from university of technology, Iraq. She is an assistant professor since 2006. Her field of interest is power system operation and control. She had more than 35 published papers. Supervised more than 30 Doctor and master students.
E-mail: 30237@uotechnology.edu.iq
Shatha S. Abdulla received her B.Sc. and M.Sc. degree in Electrical and Electronic Engineering from University of Technology, Baghdad-Iraq, in 2000 and 2003 respectively. She was awarded a PhD degree from University of Cranefield, UK, in 2018. since 2005 she has been a Lecture in Electrical Engineering in the Department of Engineering at University of Technology, where she teaches Electrical Machines, Power System, Engineering Analysis and electric circuits. E-mail 30070@uotechnology.edu.iq

Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 99 NR 4/2023. doi:10.15199/48.2023.04.29

Comparison of Thresholding Algorithms for Automatic Overhead Line Detection Procedure

Published by Paweł KOWALSKI, Robert SMYK, Gda ´nsk University of Technology

Abstract. The article presents an overview of the thresholding algorithms. It compares the algorithms proposed by Pun, Kittler, Niblack, Huang, Rosenfeld, Remesh, Lloyd, Riddler, Otsu, Yanni, Kapur and Jawahar. Additionally, it was tested how the tuning of the Pun, Jawahar and Niblack methods affects the thresholding efficiency and proposed a combination of the Pun algorithm with a priori algorithm. All presented algorithms have been implemented and tested for effectiveness in detecting overhead lines.

Streszczenie. W artykule przedstawiono przegl ˛ad algorytmów progowania. Porównano w nim algorytmy zaproponowane przez Puna, Kittlera, Niblacka, Huanga, Rosenfelda, Remesha, Lloydaa, Riddlera, Otsu, Yanni, Kapura i Jawahara. Dodatkowo przetestowano, jak dostrajanie metod Puna, Jawahara i Niblacka wpływa na skuteczno ´s ´c progowania oraz zaproponowano poł ˛aczenie algorytmu Puna z algorytmem a priori. Wszystkie przedstawione algorytmy zostały zaimplementowane i przetestowane pod k ˛atem skuteczno ´sci w wykrywaniu linii napowietrznych

Keywords: thresholding, power lines, image processing
Słowa kluczowe: progowanie, linie elektroenergetyczne, przetwarzanie obrazu

Introduction

Modern energy infrastructure is a very complex system prone to failures that should be detected and removed as soon as possible. The construction of the entire infrastructure is characterized by the presence of concentrated elements, e.g. power plants or switching stations. A significant part of the discussed infrastructure are power junkies and transmission lines which is distributed. A critical problem is monitoring and maintaining the efficiency of transmission lines [1, 2, 3]. Technological progress forces the automatic control of monitoring processes. Therefore, techniques for assessing the efficiency of transmission lines based on image processing are implemented. The image can be acquired with a camera installed in an unmanned aerial vehicle (UAV) moving along the line. The problem discussed in this paper concerns image processing to detect power cables in the video stream. The scheme for extracting objects when processing an image frame is often similar. In the initial stage, called preprocessing, edge operators and semgentation algorithms are used. Processing efficiency at this stage is crucial and has a general impact on the efficiency of subsequent processes such as object detection.

This paper focuses on the analytical evaluation of thresholding methods in terms of their application in the monitoring system of power lines from a mobile carrier. The prototype of such a system was elaborated and implemented as part of previously research [4, 5, 6, 7]. One of the important research stages was the validation of thresholding methods as shown in this study. The first section presents the binarization. The next presents the measures used in the threshold selection algorithms. These measures are used in the formulas presenting the operation of individual algorithms in the third chapter. The fourth section presents the procedure for determining the effectiveness of the algorithm. And in the section five the results are discussed.

The basis of image binarization

Edge extraction is the first step in the procedure of detecting the objects in an image. Convolution with edge detection operator results in image Z, where pixel z contains the gradient value. To reduce the amount of data and extract the edges, the segmentation is performed. This procedure relies on dividing an image into disjoint regions characterized by the uniformity of pixel values. The binarization performed as a thresholding procedure is a simplest variant of segmentation. For a given threshold, the pixels below the threshold are considered to form an object (edges) in the image, while the remaining pixels constitute the background. The thresholding algorithm is as follows

.

where T indicates the selected threshold value, z represents a pixel of thresholded image, and z is the pixel value of the image resulting from the thresholding, where 1 represents edge and 0 background. (1a) specifies a thresholding without recognizing gradient direction, (1b) specifies a thresholding extracting edges with positive gradient, and (1c) specifies a thresholding extracting edges with negative gradient. The most of popular methods of selecting a threshold value are using a histogram that represents the brightness distribution of the pixels in an image. A histogram is defined as a set of discrete values H = [h0, h1, h2, …, hn]. The individual values of hi are determined by

.

where hZi is a single element of the histogram HZ, i is the index of the element, Z is the image for which the histogram is created. In practice, the value of hZi denotes the number of z pixels whose values are equal to i.

By using the historgram, it is possible to determine the probability distribution of selected pixel values in an image

.

where pi is the probability of occurrence in the image Z a pixel value of i, sizeZ is total number of pixels in the image Z, and hi is the number of pixels with a value i. We founded, that as a result of image segmentation, the background class and the graphical object class are obtained. A known method of determining the probability of occurrence of a pixel belonging to one of the mentioned classes may be used

.
.

The probability of a pixel belonging to the background class PrZ0 (T) (4a) is the sum of the probabilities of all pixels belonging to the background class. Likewise, the probability of a pixel belonging to an object class PrZ1 (T) (4b) is the sum of the probabilities of all pixels belonging to the object class.

Another measures are the arithmetic mean of the pixel values of an image SrZ, average value of pixels belonging to the background class SrZ0 (T) and to the class of the object SrZ1 (T)

.

where sizeZ0 (T) and sizeZ1 (T) denotes the number of pixels belonging to the background class and the object class in Z, respectively. A values of the SrZ (5a), SrZ0 (T) (5b) and SrZ1 (T) (5c) can be determined directly from the image Z or using a histogram H.

Using the measures formulated above we get the variances (6) that are determined by

.

where σ2Z (6a) is the variance of image Z, σ2Z,0(T) (6b) is the variance of the background class and σ2Z,1(T) (6c) is the variance of the object class. Additional measure used in the thresholding algorithms is Shannon entropy Sh(pr) [8], which is identified by

.

where pr denotes probability. In the case of splitting the image into two classes, pr and 1 − pr denotes the probability of the pixels belonging to one class or the other, respectively. Additionally, the course of the function Sh(pr) is monotonically increasing in (0; 0, 5) and decreasing monotonically in (0, 5; 1).

Review of the threshold selection methods

This section discusses the threshold selection algorithms used in the later experimental work involving the development of a complete method for the detection of power line conductors. Using a priori knowledge in the thresholding algorithm requires the size of the object estimation. For estimation of the number of pixels representing the shape of power conductors we assumed a horizontal orientation. This means that the pixels of such an line lie on all the columns of the image. Additionally, the shape of power cables is composed of two edges, each one pixel wide. The wire in the image has an opposite gradient at the top and bottom. We can estimate the number of edge pixels using a formula

.

where cc denotes the number of wires visible in the image Z, width(Z) is the width of this image, ep and en is the number of edge pixels with positive and negative gradients, respectively and enp is the estimated number of all edge pixels in the image Z. The a priori algorithm works by setting the threshold T as high as possible, but in such a way that the number of pixels assigned to the object class is greater than or equal to the estimated number of edge pixels. Another method was proposed by Niblack [9], where the threshold value is determined as the sum of the mean SrZ and the square root of the variance σ2Z multiplied by the coefficient Nibk

.

where Nibk is a constant value coefficient Nibk = −0, 2.

Among the threshold selection algorithms, there are a number of iterative algorithms, where next value of the threshold Tn+1 using the current threshold value Tn is calculated in repeated iterations. The procedure can be carried out until the threshold value stabilizes at the desired level E, which is an absolute error |Tn+1 − Tn| < E. However, there may be cases where the threshold value will oscillate between a few close values and consequently the average of these values is determined.

The algorithm starts with a fixed initial value T0, which should be selected. Thus, initiating the algorithm with a start value close to the final threshold gives a small number of iterations necessary to stabilize the threshold value.

One of the popular iterative methods was developed by Riddler and Calvard [10]. The algorithm should be initialized with a threshold value representing the background, which corresponds to T0 = 1. If the object occupies less than half of the image the initial threshold value has to be calculated as the average value of all image pixels (10).

.

The threshold update is carried out by (11)

.

where Tn is in the current and Tn+1 in the next step threshold value.

The Lloyd method [11] is another representative of the iterative approach. The calculations are performed by

.

The variables in the relationship were described earlier. Another method used in the work is the iterative algorithm proposed by Yanni [12]. The initial threshold value T0 is

.

where minZ denotes the minimum value of the pixel present in the image Z, and maxZ maximum value. Each iteration in the discussed approach is carried out by

.

where minH0 (Tn) is the id of minimum histogram value belonging to the background class, maxH0 (Tn) is the id of maximum histogram value of the background class, minH1 (Tn) is the id of minimum histogram value of the object, and maxH 1 (Tn) specifies the id of maximum histogram value of an object class in the image Z. It has been noticed that in the case of power cables detection, in most cases the algorithm termination condition is reached after three iterations. The algorithm mentioned above has a low computational cost because the threshold is determined using the average of four values.

The Rosenfelda method [13] is based on the geometric analysis of the histogram. The histogram H can be represented graphically as a Cartesian bar graph (Fig. 1), where the tops of the bars have coordinates (i, hi). The first step in this method relies on finding the shortest polyline Ħ that lies on above the histogram H and connects its extreme ends. The shape of the Ħ polyline tracing the histogram can be written as a set of vertices. Next, the place where the H histogram is the farthest from the polyline Ħ is sought.

.

In (15) a vector of example values representing the histogram is presented. For the histogram H (15), a polyline Ħ (16) was created

.

where each point represents the vertex of the polyline Ħ. The graphic representation was shown on (Fig. 1). The final

Fig. 1. Visualization of the Rosenfeld method

step in Rosenfeld’s algorithm is to find the bar of H that is furthest from the polyline Ħ, which is h12 visible on Fig. 1.

A separate group of threshold selection methods used here are optimization algorithms. They work by finding the minimum or maximum of the goal function. One of such methods is the Otsu algorithm [14, 15], in which the selection of the threshold for dividing the image into two classes is carried out by minimizing the variance for each class. The algorithm works by selecting a threshold with the lowest intra-class variance σ2wew

.

Determining the intra-class variance is difficult. The formula for the inter-class variance σZmie(T) is simpler

.

The relationship between the global variance σ2Z, the inter-class variance σZmie(T) (18), and the intra-class variance σ2wew(T) (17) is as follows

.

where σ2Z (19) is a constant value, independent of the selected threshold T. It can be seen that minimizing the intra-class variance is equivalent to maximizing the inter-class variance. Calculating the inter-class variance is less computationally complex and is faster. Therefore, the threshold should be calculated by maximizing the inter-class variance. The Huang and Wang algorithm [16] is another thresholding method from the optimization group. It is based on the use of the membership function HWμ(z) represented by

.

The value of the function HWμ(z) for any pixel z should be within the range from 0.5 to 1. The coefficient value HWC is carried out using

.

where max(Z) is the maximum pixel value in the image Z, and min(Z) the minimum pixel value present in the image Z. The threshold value in Huang’s method is determined by minimizing the objective function Huang(T)

.

Jawahar, Biswas and Ray in [17] presented a method in which the threshold is selected by minimizing the mean square objective function Jaw(T) expressed as

.

where Jawτ denotes the amount of blur between the background and object classes in the image Z. This value has a key influence on the effectiveness of thresholding. According to [17] this argument should be Jawτ ≥ 1. The value of Jawτ = 1 corresponds to the k-means algorithm [15].

Kittler and Illingworth [18, 19] presented a method in which the threshold is selected by minimizing the function Kittler(T) expressed by (24)

.

By using the function Sh shown in (7) the equation (24) can be written as (25):

.

The method proposed by Remesh [20] is based on the minimization of the function Remesh(T) (26)

.

On the other hand, Kapur presented a method [21] in which the threshold is selected by maximizing the function Kapur(T) (27)

.

Pun [22] presented another optimization method. Using the formula developed by Shannon [8] he defined four entropies, and using them he defined the function PunFe(P unα) (28)

.

Pun in [22] proposed to search for a threshold T, by minimizing the function PunFe(P unα) On the other hand, Kapur in [21] proposed to determine the threshold T by maximizing the function PunFe(P unα).

In another work, Pun [23] proposed to search for the optimal threshold using an anisotropic coefficient, hereinafter referred to as Punβ. To determine this coefficient, it is necessary to estimate the smallest threshold T0, which divides the histogram into two parts, the first of which should contain at least half the pixels. The initial threshold may be estimated as

.

and next T0 is used to calculate the anisotropic coefficient Punβ (30)

.

Then the coefficient Punβ is used to determine the threshold value using the relationship

.

In the case of Punβ > 1/2 a probability PrZ0 (T) equal Punβ and in the case of Punβ ≤ 1/2 the provability PrZ0 (T) should equal 1 −Punβ.

Both Pun algorithms showed low efficiency in detecting the edges of overhead wires. This is because they are specialized in thresholding images to recognize text, where typically the pixel count of the object class is close to the pixel count of the background class. In the case of overhead lines, the difference in the number of pixels of each class is significant. Typically, the number of pixels representing wire edges does not exceed 1% of total pixels.

In order to adapt the second Pun algorithm to detect power line conductors, the formula (29) has been replaced by the formula

.

where the factor Punmod was introduced, which specifies the initial ratio of the number of pixels of the object class to the number of pixels of the background class.

The next modification is to determine the initial threshold T0 using the formula (8a) from the a priori method. Then Punβ is determined by using (30) and tuned up by coefficient apβ. The selection of the T treshold is done by minimizing the function apPun(T)

.
Effectiveness of thresholding algorithms comparison

The effectiveness analysis of the threshold selection algorithms was made using a representative set of photos of overhead wires. The pixels representing the wire in each photo were predetermined. The illustration of original picture and selected pixels are on Fig. 2. The function efi(x) has been defined and using the following formula for the effectiveness of detection of the i-th conductors in the column x was established

Fig.2. The test image (left) and arbitrarily selected wire pixels (right)

.

where, ~Wix stands for the set of pixels representing the i-th wire in the pixel column x, and z’x,y is the (x,y) pixel value of the image after edge extraction. The measure of effectiveness effbw is formulated as the sum of the measures efi(x) for all lines in all columns of the image

.

where n determines the number of visible wires, and width(Z’) is the image Z width. Additionally, the relative efficiency effww expressed by (36) determines the percentage number of detected columns

.

After the image has been processed using the edge detection and thresholding algorithm, the pixel counts can be determined as follows

.

where edgesPix (37a) is the total number of pixels detected as the edge, and trueEdges (37b) the number of edge pixels that represent a correctly detected wire edge. With the use of the presented dependencies on the number of pixels, the coefficient trueEdgewp (38) was formulated. It denoting the number of correctly detected edge pixels in relation to all detected edge pixels

.

The coefficients were used to elaborate a general measure of the algorithm’s effectiveness in wire detection oeff, which is expressed as

.

In the initial phase of the research on the development of a complete overhead cable detection procedure, the algorithms discussed in this paper were implemented and validated. First the software environment was used for verification of all the methods, and next the most effective methods are implemented in hardware FPGA environment. The validation was carried out using a set of photos of overhead lines taking into account the typical range of possible combinations of the location of the objects in the image against the environmental background.

Results discussion

The presented algorithms are implemented and used in the image processing procedure. It consists of image conversion to grayscale, convolution with the horizontal mask of Sobel operator [24, 4], selection of the threshold with the selected method and binarization. Then, the detection efficiency of overhead wires using this procedure was determined as oeff (39).

Several of the presented threshold selection algorithms have coefficients whose value can be fine tuned. These are the Jawahar algorithm, the second Pun algorithm with modifications, and the Niblack algorithm. The tuning of the values of the coefficients contained therein on the effectiveness of wire detection has been tested. The tests were performed using the thresholds (1a), (1b) and (1c) for the given image sets, respectively OPKnp, OPKp and OPKn.

The results of the Jawahar method effectiveness tests are presented on Fig. 3. The influence of the coefficient Jawτ on oeff was shown. The maximum effectiveness of oeffavg = 63, 5% of the Jawahar method is obtained for Jawτ = 0, 13. The effectiveness above 50% is reached forJawτ ∈ (0, 01; 1).

Fig.3. The dependence of τ on oeff in Jawahar method

Fig.4. The dependence of oeff on Punmod in the second Pun method

For the second Pun algorithm, the influence of the coefficient Punmod on the effectiveness of thresholding oeff was analyzed and shown on Fig. 4. Thresholding with Punmod > 0, 005 have the efficiency oeff < 25%. After choosing the appropriate value Punmod a significant improvement is obtained over the classical implementation of the second Pun method. Highest efficiency is achieved by modification Punmod (32) on the level of oeffavg = 65% for Punmod = 0, 00029. Modification of the coefficient Punmod by setting a constant value close to 0, increased the effectiveness of thresholding compared to the unmodified version of the algorithm.

On a Fig. 5 a results of thresholding effectiveness tests using of the second Pun algorithm combined with a priori was shown. It should be noted, that the effectiveness of the thresholding oeff is dependent on the value of the coefficient apβ. Fig. 5 illustrate the results showing the dependence of the apβ coefficient on the effectiveness of thresholding oeff. For apβ ∈ (0, 16; 0, 46) combination of a priori and Pun2 method reaches the effectiveness oeffavg above 60%. The maximum value of oeffavg = 70% is achieved for apβ = 0, 26.

Fig.5. The dependence of oeff on apβ combining the second Pun and the a priori algorithm

Fig.6. The dependence of oeff on Nibk in the Niblack method

Fig.7. Effectiveness of selected thresholding methods

In the classic version of the Niblack method used for the segmentation of images containing text, the coefficient Nibk is a constant −0, 2 [9]. The influence of modifying Nibk for effectiveness oeff of Niblack algorithm in detection of overhead wire edges was checked. The results of the tests are on Fig. 6, where the effectiveness of wire detection oeff (39) depending on the value of Nibk was shown. The effectiveness oeffavg above 60% is achieved for Nibk ∈ (5, 9; 8, 5). However, the maximum overall efficiency of the algorithm oeffavg = 66%, for the test set is obtained with Nibk = 7, 2. In order to check the maximum possible effectiveness, an analysis of the photos used to test the thresholding algorithms was carried out. The experiment consisted in checking all possible thresholds for each photo. For each image the threshold was individually selected to meet maximum oeff. The oriented algorithm is based on expert knowledge, so it cannot be implemented in a real image processing system. After analyzing the entire set of photos for that case, the maximum possible effectiveness of the threshold was determined on oeffavg = 74, 5%.

The overall test results are shown on Fig. 7 as an average measure of the effectiveness of wire detection oeffavg for each of the tested algorithms. The effectiveness of the tested algorithms has the values from 4% to 66%, and about half of the tested algorithms achieve effectiveness over 50%. The three most effective algorithms among the tested are modified by authors versions of known methods, which were not intended for wire detection. In this case the classic versions achieve efficiency below 10%. Summing up, the validated algorithms was divided into three basic groups:

• optimization — their point is to minimize or maximize the appropriate function. Its argument is usually the T threshold. These algorithms include: algorithms Jawahar, Kapur, Otsu, Huang, Kittler, Remesh, Pun1 and Pun2,

• iterative — their operation is based on a repeated calculations. The experiments shows that these algorithms usually require several steps to achieve the satisfactionary result. Representatives of such algorithms are Yanni, Riddler oraz Lloyd,

• algorithms running in place — in which the calculation procedure is one-shot, they include algorithms marked as Niblack and a priori

The most computationally expensive and time consuming are optimization algorithms. Iterative algorithms have less computational effort due to the relatively small number of steps. It should be mentioned that, for example, in processing a video stream, the threshold calculated in the previous frame can be passed as the initial value of T0 in present processed frame. In this case, stabilization usually does not exceed three iterations. The fastest and the most predictable in terms of runtime are one-shot algorithms. It is a tuned version of the Niblack method and the a priori method, which achieved efficiency of 66% and 61% respectively. In this case, the distribution of pixel values in the image does not noticeably affect the execution speed of the algorithm. The presented advantages cause that, these algorithms are suitable for transfer to the FPGA environment.

Conclusion

In the paper an analysis of the threshold selection algorithms intended for image segmentation was shown. These algorithms were appraised for their usability in detecting overhead lines. The selected algorithms have been implemented in a mobile non-contact monitoring system. Image processing and wire detection are a key component of this system. The analysis of the selection of the threshold value carried out in this work and the presented measures of effectiveness were used in the development of the mentioned monitoring system.

The performance of cable detection was compared using the measure oeff prepared by the authors. Among the tested algorithms, optimization, iterative and in-place algorithms can be distinguished. A total of 17 threshold selection methods were tested. In addition, an expert-based algorithm was developed that determined the maximum cable detection efficiency of 75%. The highest efficiency 66% was achieved by two methods, the Niblack and the second Pun method combined with the a priori method. The Pun method is an optimization algorithm. Its implementation requires a high computational effort. Whereas the Niblack method is computationally simpler because it is an in-place (one-shot) algorithm. This leads to the conclusion that this algorithm can be used in a hardware implementation of a real-time image processing system.

REFERENCES

[1] W. Doł ˛ega, “Bezpiecze ´nstwo pracy krajowych sieci dystrybucyjnych,” Przegl ˛ad Elektrotechniczny, vol. 96, pp. 21–24, 2020.
[2] W. Doł ˛ega, “Awarie krajowych linii napowietrznych–wybrane aspekty,” Przegl ˛ad Elektrotechniczny, vol. 97, pp. 9–14, 2021.
[3] T. Tambi, “Real-time monitoring of high voltage insulators in the tropical climate,” Przegl ˛ad Elektrotechniczny, vol. 96, pp. 129– 135, 2020.
[4] P. Kowalski, Application of image processing methods for overhead power lines monitoring. PhD thesis, Gdansk University of Technology, 2020. (in Polish).
[5] P. Kowalski, M. Czy ˙zak, and R. Smyk, “Comparison of edge detection algorithms for electric wire recognition,” in ITM Web of Conferences, vol. 19, p. 01044, EDP Sciences, 2018.
[6] P. Kowalski and M. Czy ˙zak, “High voltage line distance measurement and position detection based on stereoscopic image,” in 2018 International Interdisciplinary PhD Workshop (IIPhDW), pp. 54–57, May 2018.
[7] P. Kowalski and R. Smyk, “Overhead wires detection by fpga real-time image processing,” in ITM Web of Conferences, vol. 28, p. 01046, 2019.
[8] C. E. Shannon, “A mathematical theory of communication,” Bell system technical journal, vol. 27, no. 3, pp. 379–423, 1948.
[9] W. Niblack, An introduction to digital image processing. Strandberg Publishing Company, 1985.
[10] T. Ridler and S. Calvard, “Picture thresholding using an iterative selection method,” IEEE Transactions on Systems, Man, and Cybernetics, vol. 8, pp. 630–632, Aug 1978.
[11] D. Lloyd, “Automatic target classification using moment invariant of image shapes,” IDN AW126, RAE, Farnborough, Reino Unido, 1985.
[12] M. Yanni and E. Horne, “A new approach to dynamic thresholding,” in EUSIPCO’94: 9th European Conf. Sig. Process, vol. 1, pp. 34–44, 1994.
[13] A. Rosenfeld and P. De La Torre, “Histogram concavity analysis as an aid in threshold selection,” IEEE Transactions on Systems, Man, and Cybernetics, vol. SMC-13, pp. 231–235, March 1983.
[14] N. Otsu, “A threshold selection method from gray-level histograms,” IEEE Transactions on Systems, Man, and Cybernetics, vol. 9, pp. 62–66, Jan 1979.
[15] A. Wodoła ˙zski, “Wybrane algorytmy uczenia maszynowego w segmentacji obrazu kłaczków osadów ´sciekowych,” Przegl ˛ad Elektrotechniczny, vol. 97, pp. 134–136, 2021.
[16] L.-K. Huang and M.-J. J. Wang, “Image thresholding by minimizing the measures of fuzziness,” Pattern recognition, vol. 28, no. 1, pp. 41–51, 1995.
[17] C. Jawahar, P. K. Biswas, and A. Ray, “Investigations on fuzzy thresholding based on fuzzy clustering,” Pattern Recognition, vol. 30, no. 10, pp. 1605–1613, 1997.
[18] J. Kittler and J. Illingworth, “Minimum error thresholding,” Pattern recognition, vol. 19, no. 1, pp. 41-47, 1986.
[19] H. Yan, “Unified formulation of a class of image thresholding techniques,” Pattern Recognition, vol. 29, no. 12, pp. 2025–2032, 1996.
[20] N. Ramesh, J. . Yoo, and I. K. Sethi, “Thresholding based on histogram approximation,” IEE Proceedings – Vision, Image and Signal Processing, vol. 142, pp. 271–279, Oct 1995.
[21] J. N. Kapur, P. K. Sahoo, and A. K. Wong, “A new method for gray-level picture thresholding using the entropy of the histogram,” Computer vision, graphics, and image processing, vol. 29, no. 3, pp. 273–285, 1985.
[22] T. Pun, “A new method for grey-level picture thresholding using the entropy of the histogram,” Signal processing, vol. 2, no. 3, pp. 223–237, 1980.
[23] T. Pun, “Entropic thresholding, a new approach,” Computer graphics and image processing, vol. 16, no. 3, pp. 210–239, 1981.
[24] I. Sobel and G. Feldman, “An isotropic 3×3 image gradient operator. stanford artificial intelligence project (sail),” 1968.

Authors: Ph.D. Paweł Kowalski, Ph.D. Robert Smyk, Department of Electrical and High Voltage Engineering, Faculty of Electrical and Control Engineering, Gda ´nsk University of Technology, ul. Gabriela Narutowicza 11/12 80-233 Gda ´nsk, Poland, email: pawel.kowalski@pg.edu.pl, robert.smyk@pg.edu.pl

Source & Publisher Item Identifier: PRZEGL ˛ AD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 98 NR 3/2022. doi:10.15199/48.2022.03.34

Application of the Permanent Magnet Synchronous Motors for Tower Cranes

Published by Łukasz KNYPIŃSKI1,2, Jacek KRUPIŃSKI2, Poznan University of Technology(1), Krupinski Cranes Sp. z o.o.(2)

Abstract. The paper presents the results of experimental research concerning the application of the permanent magnet synchronous motors for tower cranes. The modern permanent magnet synchronous motors (PMSM) are characterized by better operating parameters in relation to squirrel cage induction machines. In the first stage of our research, the designed crane has been equipped with commercially produced permanent magnet synchronous motors. The test bench for examination of PMSM properties was built. The preliminary selected permanent magnet (PM) motors were experimentally tested to verify its performances in crane applications. Selected results of experimental research for hoist winch drive are presented and discussed.

Streszczenie. W pracy przedstawiono wyniki badań eksperymentalnych dotyczące zastosowania silników synchronicznych z magnesami trwałymi do napędu żurawi wieżowych. Współcześnie produkowane silniki synchroniczne z magnesami trwałymi charakteryzują się lepszymi parametrami eksploatacyjnymi w stosunku do silników indukcyjnych. W pierwszym etapie badań projektowany żuraw dźwigowy został wyposażony w produkowane seryjnie silniki. W celu oceny przydatności zaproponowanych silników zbudowano stanowisko pomiarowe do badania właściwości i charakterystyk. Przedstawiono i omówiono wybrane wyniki badań eksperymentalnych. Zastosowanie magnetoelektrycznych silników synchronicznych do napędu żurawi dźwigowych.

Keywords: tower cranes, permanent magnet synchronous motors, hoist winch drive, efficiency.
Słowa kluczowe: żurawie dźwigowe, magnetoelektryczne silniki synchroniczne, układ napędowy wciągarki, sprawność.

Introduction

The tower cranes are the biggest constructed cranes machines. The crane is equipped with several different motor-drives, each of with is responsible for movement in a different plane. In addition, each tower crane has other electrical motors like fans. In construction of the tower crane, the following motor-drives are used: (a) the slewing drive, (c) the trolley travelling drive and (c) the hoist winch drive. At present the three-phase squirrel cage induction motors are commonly used in construction of the tower cranes [1, 2]. In European Union the new efficiency standard (IE3) – premium efficiency standard has been introduced. Since 1 January 2017, the minimum efficiency IE3 must be maintained for induction motors with rated power 7.5 to 375 kW. Such type of motors with improved efficiency will replace the classical induction motors in modern designs of cranes.

The KR-90-5 type tower crane available in Krupinski commercial offer has an installed mechanical power of motordrives equal 25 kW [18]. The reduction of the energy consumption in the motor-drives in one crane about 5% may lead to significant energy savings. On the whole our country the electric power consumption by all tower cranes leads to a reduction of the energy resources, which also contributes to an increase on electric energy demand. Increasing the demand for electricity leads to an increase on pollution of the natural environment in Poland. This is because the major part of electric energy is produced using fossil fuels. With a slight improvement in the efficiency of a single machine, we can achieve large energy savings.

The main purpose of our research is the construction of a tower crane, which will be equipped with permanent magnet synchronous motors. The application of such type of motors is to ensure reduction of the consumption of the electric energy in designed device. The PMSM are characterized by better operating parameters in comparison to modern high-efficiency induction motors [3, 4, 5, 6, 7, 8]. The permanent magnet (PM) motors have many advantages, such as: high efficiency, high torque to mass ratio, a high power factor and wide operation range [9, 10, 11, 12, 13, 15]. In addition, the application of the permanent magnet synchronous motors will reduce the noise produced by operating tower crane. This is because such type of motor does not need a fans. In this article the application of the PM synchronous motor for hoist winch drive is presented. The presented experimental tests allow determining the efficiency for the drive system.

Requirements for crane drives

In principle, tower cranes, depending on their type, are equipped with 3 to 4 different drives. The two main drives are the hoist winch drive and slewing drive [14]. The third one is the drive for boom inclination. In cranes with the trolley jib this is the trolley travelling drive and in cranes with the luffing jib, this is the drive which changes the jib positioning angle. The fourth drive, which is presently, used more and more seldom is the crane travel drive allowing for the movement along the rail track. There is also the fifth drive in the bottom-slewing, self-erecting crane class – it is the drive of the crane setup arrangement system. In practice, this is the fourth drive as the application of the travel mechanism in the case of these cranes is marginal. The hoist drive is the drive adapted to the crane class.

The task of the hoist drive is to carry out the most optimal weight lifting and lowering operation possible within the range assumed for the crane class. The selection of such elements as the motor, gear, inverter and cable drum is determined by the necessity of achieving the greatest lifting and lowering speeds possible. The achievement of the greatest speed possible with the maximum weight and at the same time the achievement of great speeds with low weights is a real difficulty. In practice, the overall goal is to reach the best possible compromise between maximum speeds in all cases of operations under load. The ratio of the speed with the maximum weight to the maximum speed should not be worse than 1:4. Modern drives must also be provided with a possibility to reduce speed to about one tenth of the nominal speed; this is necessary during the assembly of such prefabricated elements as balconies or flights of stairs. At the same time, the function of changing the direction of rotations without closing the brake is very helpful. This eliminates small and uncontrolled changes in the position of the load during the closure and opening of the brake. The visualization of system of hoist winch drive from designed tower crane is presented in Fig. 1.

Fig.1. The visualization of hoist winch drive, 1: PM synchronous motor, 2: gear, 3: cable drum.

Slewing drive, this type of a drive determines many aspects of the crane’s operation; in principle, its characteristics affects the comfort, speed and operating speed. The crane slewing drive must be able to accelerate or decelerate the entire construction of the crane, which is characterized by the high moment of inertia. The slewing drive must also overcome the resistance related to wind, i.e. the loads with extremely variable characteristics. In such drives, the swinging of the transferred load must be compensated. The quality of operation of the slewing drive has a huge impact on the life of the load-bearing structure of the crane, especially the tower structure. The improper selection of the drive makes the operation at a higher wind speed impossible. The lack of compensation of the swinging of the lifted load makes the operation significantly longer. Depending on the operating regime, the slewing drive control system must ensure the linear dependence of the speed on the set moment.

The main task of the trolley drive is to change the lifted load radius. The load is suspended on the trolley which is connected with the drive by means of cables. The trolley is dragged to the required position. In this case the drive system and its control is less complicated than the other drives. The good drive dynamics is the most important factor. In order to allow for the compensation of the swinging load, it is necessary to ensure the power reserve of the inverter which controls the drive. The swinging of the load, especially the load with the maximum permissible weight may lead to uncontrolled rotations of the drive rotor. This risk is eliminated effectively by the application of an appropriate type of a gear.

The travel drive is the drive responsible for changing the working position of the entire crane. Currently, it is very rarely used on construction sites; it is applicable only in warehouse crane designs [19]. This is a simple type of a drive which is not provided with a complex control system. In these control systems, fluid clutches are used. Also rail brakes are important elements. The function of these brakes is to secure the position of the crane under conditions of very strong wind; they must always be applied upon completion of operations.

Drive of the crane construction system. These drives can be divided into mechanical and hydraulic drives depending on the method of erection. Describing their characteristic features must be designed for specific solutions and is not related to crane operating drives.

The experimental setup for testing PM motors properties

The assumption of the project is the construction of a technology demonstrator. The main task of the project is to perform a research and development work which aim is: (a) preparation of the experimental setup for the testing the propulsion system of a tower crane, (b) performing experimental tests of drive system components in order to verify that the drive system fulfill the technical requirements for crane applications.

The main technical assumptions consist of: (a) obtaining the assumed lifting speed at a given weight, (b) obtaining the determined speed of displacement of trolley of the crane at the given weight, (c) determination of power consumption and efficiency of the drive system, (d) determination of the noise emission level, (e) determination of the motors heating curve.

In order to experimental verification of the properties of selected permanent magnet synchronous motors, the special experimental setup was built (see Fig. 2). The computer controlled test bench enabled free shaping of the load characteristic, which is very important during testing drives systems for cranes applications.

The view of the experimental bench is shown in Fig. 3. The performed bench enables the test to all types drive-motors applied in designed tower crane. In case of the slewing drive and the trolley travelling drive the brake operation during lowering of weight have been investigated.

Fig.2. The block diagram of the experimental bench

In order to automate the measurements process, an algorithm allows saved discrete values of electromagnetic torque, phase currents, rotational speed and acceleration to CSV files was developed. Communication with the dynamometer controller is carried out using the Ethernet interface and the Modbus TPC protocol. The observations and saving of the measurement data is performed in real-time system during the algorithm work cycle.

Fig.3. The test bench with mounted PM motor

Results of measurements

The selection of the electrical components of the hoist and slewing drive and trolley drive was made based on the Beckhoff software and the given working conditions (speed, load). As part of the research work, simulation calculations were made to verify design assumptions. On the basis of results of simulation calculation and our experience the proper drives were proposed. In the table 1 the rated parameters of servo-drive for hoist winch drive are listed.

Table 1. The rated parameters of the Beckhoff AM30833T40 motor

.

In experimental tests on the suitability of selected PM motors to crane applications, a trapezoidal motion profile was used. The motion profile determines changes of the velocity during single crane’s working cycle. The motion profile is used to calculate the values of the loading mechanical torque and enables the proper selection of the motor-drive. The visualization of the courses velocity, acceleration and position during the trapezoidal profile of motion is shown in Fig. 4. In case of a trapezoidal profile, a single working cycle consists of three stages: (a) the acceleration stage, (b) the constant speed operation stage and (c) the braking stage.

Fig.4. The trapezoidal motion profile, ta is the acceleration time, tr is the time of work with constant speed, td is the braking time, tb is the break time Fig. 4. The trapezoidal motion profile, ta is the acceleration time, tr is the time of work with constant speed, td is the braking time, tb is the break time

A. Analysis for the first type of load

As a first was performed experimental test simulating weight lifting at speed equal 1200 rpm with given load torque equal 150 Nm. It was assumed that single working cycle consist working time equal two minutes and break time equal three minutes. The measured waveforms of speed and torque are presented in Fig. 5.

Additionally, the waveform of output mechanical power during enforced duty cycle with load torque equal 150 Nm has been observed. On the basis of the mechanical power waveform (Fig. 6), the average mechanical power during a single duty cycle was calculated.

Next the waveforms of supply voltages and currents of propulsion system have been measured [17]. The waveforms of three line supply voltage (u1, u2 and u3) and supply line currents (i1 and i2) are presented in Fig. 7.

On the basis of the waveforms of voltages, currents and mechanical power of the servo-drive for hoist drive the functional parameters were calculated. The values of electrical parameters are listed in Table 2.

Fig.5. The speed and torque waveforms

The efficiency of hoist winch drive system was calculated as follows:

.

where Pm is the mechanical power of hoist system, Pe is the total electrical power consumed by the hoist system.

Energy conversion efficiency for the host drive system calculated on the basis of presented measurements is equal 70.129%.

Fig.6. The output mechanical power during enforced duty cycle

Fig.7. The waveforms of supply voltages and currents

Table 2. The results of the electrical measurements for hoist winch drive system

.

B. Analysis for the second type of load

In order to accurately determine the parameters and properties of the propulsion system for hoist winch measurements for lifting and lowering the load have been performed. The following parameters have been assumed: the loading torque equal 135 Nm and rotational velocity equal 1280 rpm. In such case the single working cycle consist of: (a) two minute of working time and (b) three minute of break time. The Fig. 8 illustrates the waveform of mechanical power measured on the shaft of the motor drive.

Fig.8. The output mechanical power during second loading test

In this case, the functional parameters of the servo-drive for hoist drive are presented in the Table 3.

Table 3. The results of the electrical measurements for hoist winch drive system during second type of load

.

For second studied case, the energy conversion efficiency for the host drive system calculated on the basis electrical and mechanical parameters is equal 88.045%.

Conclusions

The paper presents an application of the permanent magnet synchronous motors as a motor-drive to tower cranes. Based on the calculation and computer simulations, the proper motors were selected to relevant systems of the hoist winch drive. In the first stage of our project, the commercially produced PM motors were applied in constructed device.

The test bench for measurement functional parameters of the selected machines was built. The test bench allowed to modeling different waveform of load torque. Using the experimental bench the functional parameters of the propulsion system of hoist winch drive have been determined.

In the future research it is planed designing a new generation of energy-saving motors for cranes applications. In optimal designing process will be applied new nondeterministic methods of optimization from group of nature inspired algorithm. Moreover the guidelines for the optimal selection of electric motors and parameterization of the power supply control will be elaborated.

Acknowledgments: This work has been financially supported by the project No. POIR.01.01.01-00-1220/17.

REFERENCES

[1] Mitrovic N., Kostic V., Petronijevic M., Jeftenic B., Multi-motor drive for crane application, Advances in Electrical and Computer Engineering, vol. 9, no. 3, pp. 57 – 62, 2009.
[2] Belmans R., Bisschots F., Trimmer R., Practical design considerations for braking problems in overhead crane drives, Annual Meetings of IEEE Industry Applications Society – IAS, vol. 1, pp. 473 -479, 1993.
[3] Knypiński Ł, Jędryczka C., Demenko A., Influence of the shape of squirrel cage bars on the dimensions of permanent magnets in an optimized line-start permanent magnet synchronous motor, COMPEL, vol. 36, no. 1, pp. 298 – 308, 2017.
[4] Zawilak J., Zawilak T., High efficiency permanent magnet synchronous motor, Przegląd Elektrotechniczny, R. 90, no. 1, pp. 224 – 226, 2014.
[5] Knypiński Ł, Nowak L., Field-circuit simulation of the dynamics of the outer rotor permanent magnet brushless DC motor”, COMPEL, vol. 30, no. 2, pp. 929 – 940, 2011.
[6] Agamloh E. B., Cavaqgninio A., High efficiency design of induction machnies for industrial applications, IEEE Workshop of Electrical Machines Design, Control and Diagnostics – WEMDCD’2013, DOI: 10.1109/WEMDCD.2013.6525163.
[7] Król E., Permanent magnet synchronous motor and induction motor – factors decreasing the efficiency (in polish), Zeszyty Problemowe – Maszyny Elektryczne, nr. 80, s. 223 – 226, 2008.
[8] Barański M., Demenko A., Łyskawiński W., Szeląg W., Finite element analysis of transient electromagnetic-thermal phenomena in a squirrel cage motor, COMPEL, vol. 30, no. 3, pp 832 – 840, 2011.
[9] Knypiński Ł., Optimal design of the rotor geometry of linestart permanent magnet synchronous motor using the bat algorithm, Open Phisycs, vol. 15, no. 1, pp. 965 – 970, 2017.
[10] Awah C. C., Okoro O. I. , Chiukuni E., Coggging torque and torque ripple analysis of the permanent magnet fluxswitching machine having two stators, Archives of Electrical Engineeering, vol. 68, no. 1, pp. 115-133, 2019.
[11] Zhang Haifeng, Dong Zhi, Zhou Jinghua, Optimization design and analysis of permanent magnet synchronous motor based on VC, The Proceedings of International Conference on Electrical Machines and Systems, DOI: 10.1109/ICEMS.2017.8055957, 2017.
[12] Sorgdrager A. J., Wang R., Grobler A. J., Multiobjective design of line-start PM motor using the taguchi method, IEEE Transactions on Industry Applications, vol. 54, no. 5, pp. 4167 – 4176, 2018.
[13] Meng Y., Meng X., Design and implementation of a PMSM servo drive systems append to intelligent patrol robots, Materials Science and Engineering, vol 397, pp. 1 – 9, doi: 10.1088/1757-899X/397/1/012064, 2018.
[14] Backstrand J. E., The application of adjustable frequency drives to electric overhead cranes, DOI: 10.1109/IAS.1992.244208, 1992.
[15] Gwoździewicz M., Zawilak J., Limitation of the torque ripple in medium power line-start permanent magnet synchronous motor, Przegląd Elektrotechniczny, R. 93, no. 6, pp. 1- 4, 2017.
[16] Geng S, Zhang Y., Qiu H., Yang C., Yi R, Influence of the harmonic voltage coupling on torque ripple of permanent magnet synchronous motor, Archives of Electrical Engineering, vol. 68, no. 2, pp.399 – 410, 2019.
[17] Bugała A., Bednarek K., The use of computer simulations and measurements in determining the energy efficiency of photovoltaic installation, ITM Web of Conferences, vol. 19, 01021, 2018.
[18] https://www.krupinskicranes.com/
[19] https://www.konecranesusa.com/industries/automotive/craneautomation-for-the-automotive-industry

Authors: dr inż. Łukasz Knypiński, Poznan University of Technology, Institute of Electrical Engineering and Electronics, Poznań, Poland, e-mail: lukasz.knypinski@put.poznan.pl, mgr inż. Jacek Krupiński, Krupinski Cranes Sp. Z o.o., ul. Obywatelska 2A, 80-259 Gdańsk, j.krupinski@krupinskicranes.com

Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 96 NR 1/2020. doi:10.15199/48.2020.01.07

Small Hydropower Plant with Variable Speed PM Generator

Published by Witold MAZGAJ, Zbigniew SZULAR, Tomasz WĘGIEL, Tadeusz SOBCZYK, Politechnika Krakowska, Instytut Elektromechanicznych Przemian Energii

Abstract. This paper presents a new concept of a Small Hydropower Plant (SHP) which is based on a permanent magnet generator (PM generator) with a propeller turbine integrated with the generator rotor. The PM generator can work at a variable speed and therefore energy produced by the PM generator has to be converted by means of a power electronic unit to fit to the three-phase power grid parameters. The paper describes elements of the energy conversion system and it also presents the results of numerical calculations of this system working.

Streszczenie. W artykule zaprezentowano nową koncepcję Małej Elektrowni Wodnej (MEW) opartej o zintegrowany z turbiną śmigłową generator synchroniczny z magnesami trwałymi. Generator pracuje ze zmienną prędkością obrotową, dlatego energia przez niego wytwarzana musi być przekształcona za pomocą układu energoelektronicznego do parametrów zgodnych z wymaganiami sieci trójfazowej. W artykule opisano elementy systemu wytwarzania i przekształcania energii oraz przedstawiono przykładowe wyniki obliczeń numerycznych pracy tego systemu. (Mała Elektrownia Wodna z generatorem z magnesami trwałymi pracującym ze zmienną prędkością obrotową).

Keywords: propeller turbine, PM generator, PWM rectifier, small hydro power plant.
Słowa kluczowe: turbina śmigłowa, generator z magnesami trwałymi, prostownik z modulacją szerokości impulsów, mała elektrownia wodna.

Introduction

Small Hydro Power Plants (SHP) are widely used across the world. Electrical generators for today’s small hydro power plants are designed for a constant rotation speed, which is kept by a speed controller often consisting of mechanical equipment. Changes of energy provided by water depend on water flow, which is very unreliable for small rivers in mountainous areas. Therefore, full efficiency can be achieved for power technology with generators working at a variable speed. So, in this paper a new solution of a PM generator integrated with a propeller turbine working at a variable speed is discussed [1].

Fig.1. Energy conversion system of the Small Hydro Power Plant

It is assumed that the mechanical system for speed control via the change of the angle of turbine blades is removed. This leads to an essential simplification of mechanical systems but this in turn requires an application of a power electronic unit (PEU) in the energy conversion system (Fig.1). The rotation speed of the propeller turbine can be variable and for different values of the water flow it should be controlled by the PEU to ensure the highest possible efficiency [2]. Due to non-linear turbine characteristics it is necessary to formulate a suitable control algorithm for the whole system of energy conversion. Therefore, the PEU has to be applied not only to ensure the output frequency and voltages required by the power grid, but also to control the energy flow from the PM generator to the three-phase grid.

This paper presents a new concept for an experimental power station with the variable speed PM generator which is integrated with the propeller turbine (nominal data of the PM generator: PN = 30 kW, UN = 500 V, IN = 34,7 A, f = 50 Hz, nN = 600 rpm).

Turbine integrated with PM generator

The hydro-set is based on a tubular turbine construction where the working fluid changes pressure when it moves through the turbine, giving up its energy. Typical solutions utilize a shaft to transfer the torque from a turbine impeller to a synchronous or induction generator.

Fig.2. View of turbine integrated with PM generator

This new construction avoids the need of a shaft and a shaft guide system. The torque is transferred by a special external ring, being an integrated part of the turbine impeller. This system is simple, durable and reliable, therefore it does not require special service. Figure 2 shows a view of the complete hydro-set designed by CEDI [3]. Permanent magnets are mounted on the external surface of the external ring [4] and spaces between magnets are filled with non-magnetic epoxy resin (Fig.3). Both the internal stator surface and the external rotor surface are protected by waterproof tubes. Water which flows through the gap between the rotor and the stator additionally ensures generator self-cooling system of stator windings and permanent magnets. Guide vanes direct the water to the turbine vanes. The water flow acts on the runner blades (Fig.3), causing runner rotation. The guide vanes can be adjusted to allow efficient turbine operation for a wide range of water flow conditions and continuous energy production with the highest possible efficiency of the whole system.

Fig.3. Location of permanent magnets on the rotor

Changes of the guide vanes angle α can be caused by hydrological conditions or by a certain decreasing of energy consumption in the three-phase grid. It requires a certain reduction of the energy production. This angle is equal to 90 degrees if the guide vanes are fully open and it equals zero when the hydro-set inlet is closed. In consequence the control system of the guide vanes should be an integral part of whole energy conversion system.

Figure 4 presents relations Pm = f(ω) between the PM generator power and the rotation speed of the hydro-set in case of constant head H (where H is a difference between upper and lower water levels) for some chosen angles of the guide vanes.

Fig.4. Relations between the PM generator power Pm and the rotation speed ω of the hydro-set for some chosen guide vanes angles α: α1 = αmax = 900, α1 > α2 > α3 for constant head

The PM generator should be specially designed for the sake of turbine dimensions. Its design should ensure water protection, proper parameters of this generator, especially its internal reactance Xd and electromotive force EMF. The mathematical model of the PM generator which assuming the base-harmonic of MMF and magnetic linearity is following:

.

The results of field computation and design formulas allowed to approximate PM flux leakage of windings and inductances for experimental PM generator:

Ls = 1,41 mH, Ms ≈ −0,5 Ls = −0,69 mH,
Lσs = 0,99 mH, Mσs = −0,32 mH,
Ψm = 1,36 Wb, Rs = 0,99 Ω
Ld = 3,42 mH, Xd = 1,07 Ω.

Water flow and energy production control system The fundamental purpose of the proposed control strategy is to transfer maximum possible amount of energy, produced by the hydro-set, to the power grid. The Energy Production Controller and Turbine Water Flow Controller (Fig.5) ensure correct operation of the whole energy conversion system during changes of working conditions.

Fig.5. Water flow and power production control system

The amount of energy, which can be produced, depends on hydrological conditions. The Turbine Water Flow Controller decides about the water stream flowing through the turbine by setting proper angle α of guide vanes on the basis upper water level UWL which should be constant.

The non-linear characteristics of the hydro-set (turbine integrated with PM generator) should be stored in a memory of the Energy Production Controller and implemented in the control algorithm. The energy which can be transferred to the three-phase grid is calculated on the basis of the turbine characteristic Pm = f(ω) (Fig.4) for the current position of the guide vanes and then, the controller determines the reference current Irs for PEU. Changes of net head additionally modify the maximum value of power on characteristic Pm = f(ω) according to following formula:

.

It is necessary to underline that the guide vanes angle α changes much slower with respect to electric quantities. So, this angle can be treated as a constant value in the PEU control system.

Power electronic unit (PEU)

The RMS voltage and the frequency of the PM generator can change about -60% ÷ +30% with respect to the nominal values. It means that the hydro-set operates at a variable speed. In the proposed energy conversion system energy produced by the generator is converted into direct current energy (DC link), and then it is transferred to power grid (3×400 V, 50 Hz) via voltage source inverter [5]. In practice, two schemes of energy conversion are used, especially with reference to wind turbine.

Fig.6. Power electronic unit with the PWM rectifier

Fig.7. Power electronic unit with the diode rectifier and the DC-DC boost converter

In the first one (Fig.6) a PWM rectifier is applied. In this case the PM generator currents have almost sinusoidal shapes. The second one (Fig.7) is based on a diode rectifier and a DC-DC boost converter which can increase voltage in the DC link. In both cases, the voltage source inverter is coupled with the power grid using a transformer or induction chokes. The second method of energy conversion is not recommended because the diode rectifier can cause significant distortion of the generator currents with respect to sinusoidal shapes. As a result of this distortion the generator torque contains an alternating component with relatively high amplitude [6]. It is a certain disadvantage because the presence of this alternating component of the generator torque can badly influence durability and reliability of the hydro-set.

PEU control strategy

The main task of the PEU control strategy is to decrease the THD factor of the current flowing to the three-phase grid. The transistors of the voltage source inverter are controlled according to the given sinusoidal current signals using hysteresis controllers. The amplitude Ism of these signals is determined by the power value which can be transferred to the three-phase grid and by the actual RMS voltage of the power grid. It is assumed that the phase shift between the given current signals and appropriate phase voltages of the three-phase grid is almost equal to zero (unity power factor). The transistors of the PWM rectifier are controlled similarly as in the PWM inverter. The amplitude IGm of the given current signals should have a value which permits to keep voltage in the DC link on the assumed level (about 800 V). These current signals are synchronized with the output voltages of the PM generator. In general, phase shifts between voltages and currents of both converters are equal to zero, although, these shifts can be changed. If the RMS voltage of the three-phase grid achieves the maximum value (for example as a result of lower consumption in the power grid) then the amplitude of the given current signals of the PWM inverter decreases. It determines a certain change of the guide vanes angle α. In this case, according to the characteristics Pm = f(ω) (Fig.4), the energy produced by the hydro-set is reduced.

Numerical calculations were made for the selected working cases, which differ due to changeable hydrological conditions or due to voltage changes in the three-phase power grid. Changes of hydrological conditions influence the rotational speed of the hydro-set. Therefore, in numerical calculations it is necessary to take into account the relation between the generator torque and its rotational speed. On the basis of the characteristics Pm = f(ω) (Fig.4) this relation can be written as follows:

.

where both Tmax and ωmax depend on the guide vanes angle α.

The chosen waveforms in the energy conversion system with the PWM rectifier in a steady state for certain working conditions are shown in Figure 8. It was assumed that the guide vanes angle α is lesser than αmax and it was equal to 30°. In this case the generator output power decreased to 9 kW. According to the characteristic Pm = f(ω), at the optimal working point, the generator rotates at 432 rpm and its voltages and currents have frequency about 36 Hz. Due to the application of the hysteresis controllers the generator and the PWM inverter currents contain certain higher harmonics, but the THD factor of these currents is lesser than 6 percent. It is worth to underline that frequencies of the transistor switching in the PWM rectifier and converter are not constant and they depend on the assumed hysteresis width. In the presented case these frequencies change between 2 kHz and 3 kHz.

Fig.8. Waveforms in the energy conversion system with the PWM rectifier in steady state for α = 30°: uG, iG, Te – voltage, current and torque of the PM generator, uinv, iinv – voltage and current of the PWM voltage source inverter

Figure 9 presents chosen waveforms when the guide vanes angle α is being changed from 90 to about 45 degrees. In order to show this energy conversion system working it was assumed that the hydro-set moment of inertia is 15 times lesser than the real value.

Fig.9. Waveforms in the energy conversion system with the PWM rectifier during the change of the guide vanes angle α: ω – generator rotation speed, uDC –voltage in the DC link, Pout –power given to the three-phase grid

The maximum energy amount produced by the PM generator decreases if the guide vanes angle α is lesser than 90 degrees according to the characteristics Pm = f(ω) (Fig.4). A short increasing of the power given to the three-phase grid (short increasing of currents) causes that the generator achieves the new rotation speed value quickly. In the steady state in new working conditions (the guide vanes angle α is lesser than 90 degrees) the current amplitudes are lesser than previously. If the guide vanes are full open again (α is equal to 90 degrees) then a short limiting of power given to the three-phase grid causes that the generator rotation speed increases quickly. It is necessary to stress that waveform shapes depend on controller settings in the energy conversion system.

The voltage of the three-phase grid can change significantly faster than the guide vanes angle. Very often an increase of the grid voltage occurs when a high-power load is turned off (power consumption in the three-phase grid is lesser than usually). Figure 10 presents the chosen waveforms for this working case. If the grid voltage achieves a given maximum admissible value then the amplitude of the PWM inverter current has to be reduced. Consequently, the current amplitude of the PM generator has to be decreased. At the same time the guide vanes angle α should be suitably adjusted, due to the significant decrease of the energy transferred to the grid, otherwise the generator speed can rise significantly.

Fig.10. Waveforms in the energy conversion system with the PWM rectifier during changes of power consumption in the three-phase grid

It was earlier remarked that energy produced by the PM generator can be converted in the power electronic system with the diode rectifier and the DC-DC boost converter. Figure 11 presents chosen waveforms in this energy conversion system for the same guide vanes angle α as in Figure 8. The generator current has non-sinusoidal shape. It is the reason that the generator torque includes an alternating component, which has significantly higher value in respect to the energy conversion system with the PWM rectifier. It can significantly lower durability and reliability of the hydro-set.

Fig.11. Waveforms in the energy conversion system with the diode rectifier in steady state: uDCin, uDCout – input and output voltage in the DC link respectively, iLDC – current of choke LDC

Fig.12. Waveforms in the energy conversion system with the diode rectifier during the change of the guide vanes angle α

During changes of working conditions the energy conversion system with the diode rectifier operates similarly to the system with PWM rectifier. Figure 12 presents chosen waveforms in the energy conversion system with the diode rectifier and the DC-DC boost converter when the guide vanes angle α is being changed from 90 to about 45 degrees. Similar waveforms in the system with PWM rectifier for the same working conditions are shown in Figure 9. The output voltage of the diode rectifier is increased to the assumed input value of the PWM inverter. It is necessary to stress once again that the generator torque contains a significant alternating component.

Conclusions

The described hydro-set has some advantages in comparison to classical hydro-generators with Kaplan turbines. The mechanical part of that hydro-set is essentially simplified because a shaft does not exist, a transmission gear is not necessary and a blades position control system is eliminated. All these facts have significant influence on investment costs and reliability of the whole Small Hydro Power plant.

Unlike energy conversion systems in windmill plants the control algorithm presented in this paper is relatively simple. The generator and PWM inverter currents are directly shaped by means of hysteresis controllers. Thank to this, the THD factor of these currents is satisfactorily low. The proposed control algorithm ensures correct operation of the whole energy conversion system in different working conditions.

REFERENCES

[1] Binder A. , Schneider T. (2005), “Permanent magnet synchronous generators for regenerative energy conversion – a survey”, 11th European Conference on Power Electronics and Applications proceedings, Dresden.
[2] Koczara W., Chlodnicki Z., Ernest E., Krasnodebski A., Sel i ga R., Brown N.L., Kaminski B, Al -Tayie J . (2008), “Theory of the adjustable speed generation systems”, COMPEL: The International Journal for Computation and Mathematics in Electrical and Electronic Engineering, Volume: 27, Issue: 5, pp.1162-1177.
[3] Norway Patent No 323150 – Integrert vannturbin og generator uten nav, owner – TURBINOVA AS, designer CEDI sp. z o. o. Poland
[4] EL-Refaie A.M. , Jahns T.M. (2008), “Comparison of synchronous PM machine types for wide constant-power speed range operations”, COMPEL: The International Journal for Computation and Mathematics in Electrical and Electronic Engineering, Volume: 27 Issue: 5, pp. 967-984.
[5] Baroudi J . , Dinavahi V. , Knight A. (2007), “A review of power converters topologies for wind generators”, Renewable Energy, No. 32, pp. 2369-2385.
[6] Danilevicz Y., Drozdowski P., Mazgaj W., Sobcz yk T. , Szular Z. (2005), “The influence of failures of a multiphase p.m. synchronous generator and a static voltage converter system on the generator electromagnetic torque”, PowerTech 2005 Conference proceedings, Sankt. Petersburg, paper number 520.

Autorzy: Dr inż. Witold Mazgaj, e-mail: pemazgaj@cyfronet.pl; Mgr inż. Zbigniew Szular, e-mail: aszs@poczta.fm Dr inż. Tomasz Węgiel, e-mail: pewegiel@cyfronet.pl; Prof. dr hab. inż. Tadeusz Sobczyk, e-mail: pesobczy@cyfronet.pl; Politechnika Krakowska, Wydział Inżynierii Elektrycznej i Komputerowej, Instytut Elektromechanicznych Przemian Energii, ul. Warszawska 24, 31-155 Kraków.

Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 87 NR 5/2011

Eccentricity Fault Identification in Round Rotor Synchronous Motors considering Load Variation

Published by Bashir Mahdi EBRAHIMI1, Jawad FAIZ1, Mohammad ETEMADREZAIE2, Mojtaba BABAIE3, Center of Excellence on Applied Electromagnetic Systems, School of Electrical and Computer Engineering, University of Tehran, Tehran, Iran(1), Department of electrical and computer engineering, Delft, Zuid-holland, Netherlands(2), Faculty of Engineering, Science and Research Campus, Islamic Azad University, Tehran, Iran(3)

Abstract. Impacts of load variation have not been so far investigated on the eccentricity fault diagnosis in synchronous motors. In this paper, a synchronous motor under static and dynamic eccentricities with different load levels is modeled using winding function method (WFM) and finite element method (FEM). Self and mutual inductances of the stator windings, stator currents and torque are calculated and analyzed. Spectrum of the stator currents is utilized for eccentricity fault detection, its type recognition and its degree determination. Variation of eccentricity degrees and load levels on the selected indices is scrutinized separately and simultaneously. The accuracy of the obtained simulation results is verified by FEM and experimental results.

Streszczenie. Zbadano wpływ obciążenia na błąd diagnozowania ekscentryczności w silniku synchronicznym. Modelowano silnik używając metody WFM (Winding function method) i metodę elementu skończonego. Analizowano indukcyjności własna i wzajemną stojana oraz moment. Analizę widmową prądu stojana użyto do oceny ekscentryczności. Oceniono dokładność diagnozowania. (Błąd oceny ekscentryczności w silniku synchronicznym o różnym obciążeniu)

Keywords: synchronous motor, WFM, FEM, fault diagnosis.
Słowa kluczowe: silnik synchroniczny, ekscentryczność.

Introduction

Synchronous machines are essential and valuable parts of power systems. These machines are generally well constructed and robust. However, they are subjected to a wide variety of abnormal operations due to stress involved in electromechanical energy conversion process. Operation of synchronous machines under fault conditions disturbs their performances and declares their life spans. Also, persistent faults damages these machines and consequences outage time for repairing is costly. Thus, fault detection and condition monitoring of the synchronous machines allow more flexibility in operation by knowing the performance and extend machine life by adjusting the operation to avoid known operating regimes or ranges and cost effectiveness. Consequence of many electrical and mechanical faults occurring during the operation of electrical machines is the eccentricity between the rotor and stator [1]. Eccentricity is categorized into three general groups: static eccentricity (SE), dynamic eccentricity (DE) and mixed eccentricity (ME).

In [2], two air gap search coils method applied to diagnose the static and dynamic eccentricity faults in the synchronous generators. It has been shown that the odd multiple harmonics of the fundamental frequency present at the EMF of search coils in the presence of the SE fault. It is noticeable that the stator currents and the shaft signals of the synchronous generators are two essential parameters that so far have been used to analyze synchronous machines under eccentricity fault. Unbalanced flux-linkages on the shaft caused by the asymmetrical condition can generate shaft signals. Variations of magnetic properties, tolerances in physical dimensions and any asymmetrical induced currents in the winding or stray paths may cause unbalanced flow of fluxes around the shaft [3, 4]. Also, sectional frames and segmental punching are the major cause of asymmetries of synchronous machine [5]. The rotor eccentricity is one of the prominent causes of magnetic asymmetries and shaft voltage. It has been shown that the shaft signals reduce in tilted rotor condition due to the opposite shaft flux- linkage at both ends. Also, the amplitude of the shaft signal is proportional with the eccentricities. Shaft signals of large turbo generator have been used to detect the faults [6, 7]. The shaft signals and other related methods have been frequently used for the eccentricities diagnosis of asynchronous machine [8, 9].

The measured stator currents are utilized for signature analysis in condition monitoring [10]. So far static and mixed eccentricity faults in synchronous machines have not been diagnosed using the analysis of the stator current signature. However, analysis and diagnosis of dynamic eccentricity faults using stator current signature analysis has been reported in [10, 11]. The amplitude of the 17th and 19th harmonic components as an index was utilized for eccentricity fault recognition in synchronous generators. Stator and rotor windings distribution and air-gap permeance have been taken into account in [12] while this was not considered in [10, 11]. In [13], effects of different degrees of static eccentricity (SE) and dynamic eccentricity (DE) on the synchronous generator inductances have been investigated. However, influence of load variation on the fault detection procedure has not been investigated in [13].

Eccentricity fault diagnosis in synchronous machines is subject to consider faulty generators under different conditions. These conditions are variation of load levels, and variation of static and dynamic eccentricities degrees. Impacts of these states on the nominated indices should be analyzed to evaluate indices efficiency for incisive fault detection. According to [1]-[13], albeit eccentricity fault identification has been illustrated, the eccentricity severity has not been determined in faulty synchronous machines. Furthermore, effects of load variation on the nominated indices (17th and 19th harmonic components) have not been studied.

In this paper, a synchronous motor under different degrees of SE and DE is modeled by winding function method (WFM) and finite element method (FEM). Stator currents are calculated using these approaches for processing and feature extracting. Then, spectrum of the simulated currents is evaluated to estimate the degree of eccentricity. After that, competency of the aforementioned indices is evaluated in different load levels. Finally, simulated results are validated by experimental results.

Modeling of Synchronous Motors using WFM

Winding function theory facilitates the performance analysis of electrical machines under different internal faults such as eccentricity between stator and rotor. The major part of this theory is calculation of machine inductances considering time and spatial harmonics; the first step in the evaluation of the machine inductances is determination of the magnetic permeance distribution of the air gap in the eccentricity fault condition. This distribution has an inverse relationship with the air gap length and a direct relationship with the radius. The air gap length and radius are precisely calculated in this paper. The next step is the use of Ampere’s law on a closed-path, the result defines a turn function of the motor circuits and calculates the field intensity in the air gap. The resulting field intensity depends on the air gap magnetic permeance distribution and magnetic field intensity at the origin. Application of Ampere and Gauss’s laws lead to an expression for the magnetic field intensity of the winding, which is proportional to the magnetic permeance distribution. The proportionality factor is called the winding function of the specific winding. Having the magnetic field intensity from any winding and applying the electromagnetic laws, an integral form equation is obtained for different inductances of the motor. By applying the Ampere’s law and considering the turn function [10, 11], the following equation for winding is obtained as follows:

.

Since the iron permeability, µfe , is considerably larger than 1, the mmf drop within the iron can be ignored, compared to that of the air gap. Also, because the air gap length is small, it is assumed that the magnetic field intensity within angle φ is independent of the radius and its value is equal to that at the middle of the air gap. The inductance between two arbitrary windings is:

.

where p(φ) is the air gap magnetic permeance distribution, nH(φ) is the turn function of winding H, and NK(φ) is the winding function of winding K. The last step entails computation of the motor inductances. The calculated inductance matrices lead to a quick solution of the differential equations. If such analytical expressions are not used, all inductances profiles of the motor must be calculated at several rotor angular positions and stored in the inductances matrices. For solving it, an interpolation is then carried out for all inductances. This results in a longer computation time and lower computations accuracy. The method presented here is based on a manual computation of all inductances and deriving their closed-form analytical equations. Computerizing this method leads to the development of a software function that calculates the inductance matrix of the motor by receiving the mechanical position of the rotor and geometry of the eccentricity condition. After calculation of self- and mutual-inductances of the healthy and faulty motor for several rotor angular positions and storing data in the inductances matrices, computation of stator currents, torque and speed are necessary for processing and feature extracting. Hence the electromagnetic coupling model of the synchronous motor is solved using 4th- and 5th- order Runge-Kutta approach. Fig.1 depicts the per phase self-inductance of the stator winding of a synchronous motor in healthy, 10% DE and 40% DE cases. It is seen the dynamic eccentricity increases the magnitude of the self-inductance of the stator winding. As shown in Fig. 1, the high level DE distorts the self-inductance of the stator winding. The reason is that in the dynamic eccentricity case, the air gap permeance depends on the rotor angular position and this angle varies continuously.

Fig.1. Per phase self- inductance of stator winding for healthy and different faulty cases

Fig.2. Flux distribution in faulty motor cross-section at no load and rated excitation current

Modeling Synchronous Motors by TSFEM

According to [14, 15], the model of the faulty motors is the first stage of any reliable fault recognition algorithm. Thus, practical conditions of the faulty motor should be considered for accurate modeling. The modeling methods which are based on the magnetic field determination and also they consider different aspects of faulty machines can be selected as an efficacious approach to calculate required signals and parameters for processing. In this paper, the healthy and faulty synchronous motors under SE and DE are modeled using TSFEM. In this modeling, geometry of the motor elements including stator, rotor and shaft are taken in to account. Moreover, spatial distribution of the stator windings, non-uniform air-gap, physical conditions of the stator conductors, rotor, shaft and air-gap, and nonlinearity of the core materials are taken into account. Three-phase sinusoidal voltage applied to the terminals of the motor is the input of the simulation procedure. Specifications of the proposed synchronous motor have been given in Table I.

In this simulation, transient analysis of rotating machines is employed for modeling and analyzing the synchronous motor with mechanical coupling. The electrical equations due to the external circuits which exhibit supply and electrical circuits are combined with magnetic field equation in FEM and motion equations of the mechanical coupling.

Fig. 2 illustrates flux distribution in the simulated synchronous motor. As shown in Fig. 2 the eccentricity clearly affects the flux distribution within the round-rotor motor and the magnetic flux distributions are not identical on both sides of the motor in faulty condition. The reason is that the flux path reluctance in this type of motor is mainly determined by the length of the air gap. When the rotor is displaced to the stator, the length of gap between rotor poles and stator core in one ha lf of the rotor surrounding air is reduced, while rising in the other half. This leads to the creation of unbalanced magnetic reluctance paths for the flux and consequence asymmetry in flux distribution within motor. The eccentricity is too capable to more clearly affect to the total reluctance of the motors in low excitation current, in which the motors magnetic circuit are linear and saturation not occurred. Therefore, the full reluctance paths of the motors dominantly are determined by the air gap reluctance.

Fig.3. Per phase mutual- inductance of stator winding for healthy and different faulty cases

Table I. Salient Pole Synchronous Generator Parameters Used in the Simulation

.

Calculation of winding inductances is a prominent approach to analyze the electrical machine under different fault conditions. In order to predict performance of the motors under dynamic eccentricity, inductances of the stator circuits should be estimated precisely. Due to the symmetrical distribution of the windings in the motors slots, the inductance profile of stator windings will have identical variation with a phase shift in the presence of the eccentricity. Consequently, one phase of stator (phase A) is considered in this paper. Self inductance of phase A (Laa) is calculated from the flux-linkage seen by the phase (λaa), when a dc current passes through it, and other windings are open-circuited, as follows:

.

Fig. 3 demonstrates the self-inductance of phase A in the healthy and in case of different DE degrees in round-rotor synchronous motor. They have been obtained for different rotor angular positions from 0° to 360° with step-angle of 3°. According to Fig. 3, it is clear that the minimum and maximum amplitude of the self-inductance of the phase A rises with the increase of DE. It is seen 10% DE and 40% DE increase the self-inductance mean 4% and 12%, respectively.

Modeling Synchronous Motors by TSFEM

Following the calculation of the stator current using WFM and TSFEM, determination of the stator current spectrum is necessary to introduce a criterion for fault detection. The maximum required frequency is around 1000 Hz (21th harmonic component) and according to the Nyquist law, sampling frequency is set equal to 2000 Hz. However, in order to have an acceptable resolution, sampling frequency has been set to 6400 Hz (128 sample per cycle).

Fig.4. Normalized line current spectra of the motor using WFM in (top) healthy, (middle) 30% SE and (bottom) 50% SE

Fig.5. Amplitude variation of 19th harmonic component versus SE and DE degrees obtained by WFM.

Furthermore, the stator current signal is simulated over 4 seconds which allows analyzing the signals with a frequency resolution of 0.25 Hz. As shown in Fig. 4, 30% SE increases the amplitude of the 17th and 19th harmonic components about -2.05dB and -2.21dB, respectively. Furthermore, fault expansion to 50% SE raises the amplitude of the 17th and 19th harmonic components about 6 dB and -6.1dB, respectively. Comparison between amplitude of the 17th and 19th harmonic components in the stator current spectra due to SE reveals that incremental rate of the SE and DE degrees. Fig. 5 shows that the slope of AB is as same as BC. Indeed, incremental rate of the amplitude of the 17th harmonic component due to different SE degrees is as equal as DE cases. Therefore, it may be concluded that influence of the SE and DE fault on the stator currents harmonics is equal. It is related to the unbalanced magnetic pull (UMP) due to eccentricity which is larger for the SE cases. Fig. 6 exhibits the spectrum of the stator current which has been calculated by FEM.

Fig.6. Normalized line current spectra of the motor using FEM in (top) healthy, and (bottom) 50% DE

Fig.7. Normalized line current spectra of the motor using experiment in (top) healthy, and (bottom) 50% DE [13]

Comparison between Fig. 4, Fig. 6, and experimental results in Fig. 7 demonstrates that the obtained results by FEM is more precise than obtained results by WFM. The reason is the utilized approximation in the WFM in which non-linearity characteristics of the ferromagnetic materials, spatial distribution of the stator windings and stator slots effects are ignored.

Impacts of Load Variation on Fault Recognition

A. Stator Currents

Fig. 8 shows the variation of the amplitude of the 19th harmonic components versus different load levels. According to Fig. 8, amplitude of the aforementioned harmonics rises due to the increase of the eccentricity degree and it is fairly constant for load variation between 0% and 120% rated load. It is seen, that the load variation has no influence on the magnitude of the aforementioned harmonics. This may be justified based on the speed profile of the healthy and faulty synchronous motors. Since the synchronous motors rotate with a constant synchronous speed, load variation has no noticeable impacts on the amplitude of harmonic components at the same eccentricity degrees. Indeed, these criteria are robust against the load variation and it is not necessary to specify motor load for accurate fault detection. Therefore, these criteria are much efficient than the global indices which are utilized for the eccentricity fault diagnosis in induction motors. Amplitude of the aforementioned harmonic components has been summarized in Table II.

Fig.8. Amplitude variation of 19th harmonic component versus SE degrees and different load levels obtained by WFM

Fig.9. Torque profile of the full load synchronous motor (top) healthy and (bottom) 50% SE

B. Developed Torque

Fig. 9 depicts the time variation of the developed torque of the healthy and faulty motor for 50% and 100% rated load cases. As shown in Fig. 9, eccentricity raises the variation rate of the motor torque. This is due to the increase of magnitude of harmonic components in the stator currents and distortion of the airgap flux density. In addition, it is seen that load increase causes to decrease of the overshoot of the motor torque.

Modeling Synchronous Motors by TSFEM

In this paper, WFM and FEM were used to simulate the healthy and faulty synchronous motors under different SE and DE degrees. It was illustrated that SE and DE have the identical impacts on the inductances and stator currents spectra of the synchronous motors. In addition, it was demonstrated that load variation has no influence on the 17th and 19th harmonic components of the stator currents spectra. Indeed, these criteria are robust against load variation which is the noticeable feature for accurate fault recognition in different conditions of the motor.

Table II. Amplitude of the 17th Harmonic Components for different SE degrees and different Loads

.

REFERENCES

[1] B. M. Ebrahimi and J. Faiz, “Diagnosis and performance analysis of three-phase permanent magnet synchronous motors with static, dynamic and mixed eccentricity,” IET Electric Power Applications, vol. 4, no. 1, Feb. 2010, pp. 53- 66.
[2] Stoll, R. L., Hennache, A., “Method of detecting and modeling presence of shorted turns in DC field winding of cylindrical rotor synchronous machines using two airgap search coil,” IEE Proceeding Vol. 135, No. 6, pp. 281-294, November 1988.
[3] Hsu, J. S., and Stein, J., “Effect of eccentricities on shaft signals studied through windingless rotors,” IEEE Transaction on Energy Conversion, Vol. 9, No. 3, pp. 564:571, September 1994.
[4] Hsu, J. S., and Stein, J., “Shaft signal of salient-pole synchronous machines for eccentricity and shorted-field-coil detections,” IEEE Transaction on Energy Conversion, Vol. 9, No. 3, pp. 572-578, September 1994.
[5] Alger, P. L., and Samson, H. W., “Shaft currents in electric machines,” AIEE Transaction pp. 235-245, February 1924.
[6] Verma, S. P., and Girgis, R. S., Shaft potentials and currents in large turbo generators. Report for the Canadian Electrical Association, Research & Development, Suite 580, One Westmount Square, Monteral, Quebec, H3Z 2P9, May 1981.
[7] 61. Meyer, A., Joho, R., Posedal, Z., Reichert, K., and Ammann, C., “Shaft voltage in turbosets: Recent development of a new grounding design to improve the reliability of the bearings,” Int. Conference on large high voltage electric systems, paris, August 1988.
[8] Cameron, J. R., Thomson, W. T., and Dow, A. B., “Vibration and current monitoring for detecting airgap eccentricity in large induction motor,” IEEE Proceeding, Vol. 133, No. 3, pp. 155- 163, May 1986.
[9] Pollock, G. B., and Lyles, J. F., “Vertical hydraulic genertors experience with dynamic airgap monitoring,” IEEE Transaction on Energy Conversion, Vol. 7, No. 4, pp. 660-667, December 1992.
[10] Toliyat, H. A., and Al-Nuaim, N. A., “Simulation and detection of dynamic airgap eccentricity in salient-pole synchronous machines,” IEEE Transaction on Industry Application, Vol. 35, No. 1, pp. 86-93, January/February 1999.
[11] Al-Nuaim, N. A., and Toliyat, H. A., “A novel method for modeling dynamic air-gap eccentricity in synchronous machines based on modified winding function theory,” IEEE Transaction on Energy Conversion, Vol. 13, No. 2, pp.156- 162, June 1998.
[12] Tabatabaei, I., Faiz, J., Lesani, H., and Nabavi-Razavi, M. T., “Modeling and simulation of a salient–pole synchronous generator with dynamic eccentricity using modified winding function theory,” IEEE Transaction on Magnetics, Vol. 40, No. 3, pp.1550-1555, May 2004.
[13] J. Faiz, Bashir Mahdi Ebrahimi, M. Valavi and H. A. Toliyat “Mixed eccentricity fault diagnosis in salient-pole synchronous generator using modified winding function method,” Journal of Progress In Electromagnetics Research B, vol. 11, 2009, pp. 155–172.
[14] J. Faiz, Bashir Mahdi Ebrahimi, B. Akin, and B. Asaie, “Criterion function for broken-bar fault diagnosis in induction motor under load variation using wavelet transform,” Journal of Electromagnetics, Taylor & Francis, vol. 29, May 2009, pp. 220-234.
[15] J. Faiz and Bashir Mahdi Ebrahimi, “Determination of number of broken rotor bars and static eccentricity degree in induction motor under mixed fault,” Journal of Electromagnetics, Taylor & Francis, vol. 28, Aug 2008. pp. 433 – 449.

Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY (Electrical Review), ISSN 0033-2097, R. 87 NR 5/2011

The Optimisation of the Usage of Electricity from a Wind Turbine in a Household

Published by Wiesław MICZULSKI1, Oliwia PERS2, Uniwersytet Zielonogórski, Instytut Metrologii, Elektroniki i Informatyki (1), ETB Solution GmbH (2) ORCID: 1. 0000-0002-8175-0078; 2. 0000-0002-2962-5261

Abstract. The article presents the concept of an algorithm that optimizes energy management in a household. The algorithm optimizes the usage of the energy from a wind turbine and ensures the greatest possible comfort of using energy-efficient smart appliances by the user.

Streszczenie. W artykule przedstawiono koncepcję algorytmu optymalizującego zarządzanie energią w gospodarstwie domowym. Algorytm optymalizuje wykorzystanie energii z turbiny wiatrowej oraz zapewnia jak największy komfort korzystania z energooszczędnych inteligentnych urządzeń przez użytkownika. (Optymalizacja wykorzystania energii elektrycznej z turbiny wiatrowej w gospodarstwie domowym).

Keywords: energy management algorithm, renewable energy sources, smart appliances, wind turbine.
Słowa kluczowe: algorytm zarządzania energią, odnawialne źródła energii, inteligentne urządzenia, turbina wiatrowa

Introduction

The energy systems in many countries are developing rapidly in order to meet the increasing demand for electricity, as well as the expectations of energy consumers. Renewable energy sources (RES) are more and more often used as basic power sources also in households. The increase in renewable energy installations in Poland and Germany has occurred as a result of the implementation of appropriate financing projects for such investments. These solutions make it possible to reduce the high costs of electricity consumption from the grid. The projects promoting the financing of different types of renewable energy are being implemented in Poland and Germany.

The usage of solar and wind energy to generate electricity for households depends heavily on the weather conditions, the season and the time of day. This has a significant impact not only on the proper functioning of the electrical grid (EG) [1] but also the costs associated with energy consumption. It also has an impact on the proper management of the operation of devices in households. The work [2] presents an overview of the latest literature on energy management solutions in households.

The article presents the concept of an energy management algorithm (EMA) in a household. In the optimization process, EMA takes into account the power generated by the wind turbine (WT), which results in the reduction of energy consumption costs from EG. EMA also provides the greatest possible comfort for the user of the energy-efficient smart appliances. The choice of a WT has been determined by increased electricity production and economic considerations compared to a solar power plant [3]. The research results presented in this article are a continuation of previous research papers, the results of which are described in [4, 5].

Energy management algorithm

The goal of the EMA (Fig. 1) is to ensure that the power generated by WT (PWT) is used as much as possible to power smart household appliances. It will also reduce the cost of consumed energy from EG. In the process of optimization, EMA also provides the greatest possible comfort for the user to use smart appliances. For this purpose, it has been assumed that each of the appliances will be assigned a priority (pr) indicating which of them will be able to change power (PA) first, will be turned off or turned on with a specific time shift (ts).

Fig.1. The block diagram of an energy management algorithm (EMA) in a household

After reading the current PWT(i) value and the sum of power of the powered on appliances in a given iteration Ph(i) and the current tariff tz, the condition is checked

.

If condition (1) is not met, iteration (i) is terminated. This means that appliances that are turned on are powered only from the WT. However, failure to meet this condition will result in further EMA operation, i.e. the calculation of ∆Ph(i) according to the formula

.

In the next step of the iteration, the simultaneous fulfillment of the condition is checked

.

where: tz – tariff time zone, ∆Phys – accepted allowable value of changes |∆Ph(i)|, limiting the number of power changes of smart appliances (PA).

If condition (3) is not met, iteration (i) is terminated. However, if this condition is met, the sub-program “Optimizing the power of smart appliances” will be started, preceded by the reading of the current parameters of active appliances. For the optimization process, the GRASP (greedy randomized adaptive search procedure) heuristic algorithm [4, 6] has been used, in which the objective function takes into account: the current PWT values and power consumed from EG (PEG), current power values of individual appliances turned on (PA), remuneration (cWT) for 1 kWh transferred from WT to EG and the cost of 1 kWh from EG (cEG) determined for the relevant tariff, ts and pr values of active smart appliances and the total number of them in the household. Compared to [4], an additional condition has been introduced in the optimization process

.

where: (Ph_EMA) – value of the sum of the power of all appliances after optimization by EMA.

Condition (4) allows for a better adjustment of the power of appliances to the power generated by the WT at the expense of a small amount of energy taken from the EG, resulting from the adopted ∆P value.

As a result of the optimization process, the following decisions can be made in relation to a smart appliances:

– disagreement with its activation,
– its activation or shift of the activation time (ts),
– change of power value,
– its deactivation.

In the sub-program “Confirmation by the user” will be displayed information about the actions taken by the EMA in relation to the smart appliances. The user may knowingly allow or reject the proposed proposal. If the user does not react to the information displayed, the EMA will automatically consent to the implementation of the proposed proposal after the set time.

Characteristics of smart appliances in the household

In order to verify the functioning of the EMA algorithm, it has been assumed that the household would use energy efficient smart appliances, equipped with systems for measuring the power consumed by each appliances (PA) (e.g. [7]), and remotely controlled to change the power consumed, turn them off and time delay (ts) of their activation.

Table 1 presents a list of smart appliances installed in a sample household for which the values of PA, ts and pr have been defined. For each appliances turned off, shown in Table 1, PA = 0 kW has been assumed. The boiler has 3 heaters installed with the power of 1.5, 1.5 and 2 kW, which allow the EMA to set one of the following PA values: 1.5, 2, 3, 3.5 and 5 kW. The air conditioner also has the option to set the PA values: 2.2, 3 and 3.5 kW. Both of these appliances turn off automatically as a result of reaching the predetermined values of the output parameters.

Table 1. Parameters of smart appliances

.

*PA – the power value depends on the individual control sub-program installed in the smart appliances.
**PA – the power value depends on the EMA decision.

It has been also assumed that the EMA can turn off appliances with pr equal to 2, 3 and 5. On the other hand, appliances with pr = 6 will not be turned off by the EMA. Each appliances will be turned on after obtaining approval from the EMA or as a result of an informed decision of the user. For washing machines and dishwashers, pr = 1 has been assumed, as these are appliances whose starting time can be shifted by the EMA to a very large extent. This will have a positive effect on the optimization process. The appliances with pr = 4 will be turned off automatically as a result of completing the program in progress. It has also been assumed that the electric car battery charger will be turned on in the night time zone. The battery charging time from 10% to 100% of its capacity is 9 hours and 40 min. The first phase of battery charging is carried out with a constant current value and lasts about 1.5 hours.

In order to manage the energy in the household, it has been proposed to build an intelligent network Home Area Networks (HAN), the diagram of which is shown in Figure 2. The basic element of this network is the energy management module (EMM) consisting of, among others, from a microcontroller with an installed EMA and a system for measuring the power generated by WT. It has been assumed that smart appliances (Tab. 1) in a household would be powered by energy generated by WT with a power of 15 kW [8]. When determining the power value of the wind turbine, the car battery charger has not been taken into account. This has been based on the assumption that the electric car battery would be charged during the night tariff, which is cheaper than the daily tariff. Also at this time, there is often a small number of household appliances turned on.

In the case of a lower production or a complete deficit of energy from the WT, it will be replenished with EG. EMA, by optimizing the usage of the energy from WT, and thus the energy costs, will accordingly reduce the energy consumed by smart appliances from EG. Smart appliances will be controlled via the HAN network. The amount of energy taken from the EG, as well as the amount of energy transferred to this grid as a result of the excess energy generated by the WT, will be measured by an smart meter (SM). The communication of the SM with EMM will also be carried out by HAN.

Fig.2. The diagram of a smart grid in a household

The results of simulation tests of the energy management algorithm

The conducted simulation tests have been aimed at checking the correctness of the developed EMA concept. In addition to the above assumptions, simulation tests have been conducted under the following additional assumptions:

– the household is located in Jena (Germany), with a strictly defined Time-of-Use (ToU) pricing program with two time zones (tz) [9]:
• tz = 1 from 0:00 to 6:00 and from 22:00 to 24:00 and the price for 1 kWh of energy (cEG_1) is 0.32 €,
• tz = 2 from 6:00 to 22:00 and the price for 1 kWh of energy (cEG_2) is 0.41 €,
– a recompense for 1 kWh of energy (cWT) transferred from WT to EG is equal to 0.092 € (applies to Jena) [10],
– ∆Phys = 0.2 kW,
– ∆P = 0.4 kW.

The simulation tests have been carried out for the period ofThe simulation tests have been carried out for the period of 1 day for a household equipped with appliances presented in Table 1. The time characteristics of these appliances have been determined on the basis of a database [11]. The results assessing the benefits of using EMA and WT for five analyzed cases are presented in Table 2.

In case 2, the EMA performed an energy optimization for tz = 2 by shifting the reported washing machine operation from 7:14 per hour 22:00 (tz = 1). At 9:06 there has been a notification that the dishwasher had started working. EMA proposed to move the operation of this appliances to the zone tz = 1 per hour. 23:00. The user did not agree to the algorithm proposal, as he had another launch in the plan. After this decision, the dishwasher has been turned on and finished its work at 9:37. At 10:09 there has been another application of the dishwasher for operation. In this case, the user agreed to move the dishwasher’s operation to 23:00. In cases 1 and 2, the same amount of energy was taken from the EG by all appliances. However, in the second case, the energy costs (Tab. 2) decreased by 1.16% (0.21 €) as a result of the operation of the EMA.

In case 3, the reduction in energy costs, compared to cases 1 and 2, has been only due to energy generation by the WT. The household profit was 6.60 €.

Table 2. Results of simulation tests for five cases

.

In case 4, the EMA used reduced the amount of energy consumed from the EG for tz = 2 by adjusting the total power of the appliances included (Ph_EMA) so that it is not greater than the PWT power by the adopted value of ∆P = 0.4 kW (Fig. 3).

Included the values of the quantities defined in the objective function, the EMA performed the following sequence of actions on the air conditioner: at 6:27 he was reported to work, but due to the low value of the power generated by the WT, it has not been turned on (Fig. 3). The consent of the EMA to turn on the air conditioner has been made at 6:42, but with PA = 2.2 kW taking condition (4) into account. At 6:47 the air conditioner turned off automatically. Later in the day, the EMA did not take any action on this appliance. In the case of the boiler, EMA performed from 6:45 to 7:32 activities to optimize Ph_EMA, so that according to condition (4), the determined value ∆P with respect to PWT changes is not exceeded. The sequence of EMA activities during this time period has been as follows: at 6:45 a boiler has been reported for operation and it has not been turned on as the air conditioner has been running at that time. After the air conditioner has been turned off automatically at 6:47 EMA turned on the boiler with PA = 2 kW and turned it off at 6:50 after turning on the kettle, which could have been turned on according to condition (4). At 6:53 the EMA turned on the boiler with PA = 3 kW, and at 6:58, according to the condition (4), it reduced its power to 2 kW. Failure to take into account the condition (4) in the algorithm would cause its reaction at 6:53 to not turning on the boiler with PA = 3 kW but with PA = 2 kW. Such action would extend the operation of the boiler. At 7:09 the EMA again increased the boiler capacity to 3 kW. After turning on the washing machine at 7:15 EMA reduced the boiler power to 1.5 kW, by 7:19 increased to 2 kW; and at 7:24 turned off the boiler due to the air conditioner being turned on. After the air conditioner turned off automatically, the boiler has been turned on again at 7:29 with PA = 3 kW and turned off at 7:32 due to the lower PWT power generation with the washing machine still running and the kettle turned on. The next action of the EMA has been at 14:06. When turning on the kettle, EMA reduced the boiler power from 5 kW to 2 kW, and after turning off the kettle at 14:09 the boiler power has been increased to 5 kW. At 14:13 the boiler has been turned off by the EMA because the air conditioner has been turned on and it turned off automatically at 14:20. The EMA then turned on the boiler with PA = 5 kW again, and after 1 minute it turned off automatically. At 16:16, with the air conditioner and the EMA kettle on, he changed the boiler power from 5 kW to 3 kW for a period of 1 minute. As a result of the above EMA actions, in case 4, the final profit for the household (Tab. 2) was higher than in case 3 by 3.93% (0.27 €).

In case 5 (Tab. 2), between 6:00 and 24:00 there has been a large power generation by WT (PWT > Ph) which resulted in no EMA operation for tz = 2. Moreover, a high profit (16.35 €) has been achieved in the analyzed household.

Fig.3. Illustration of EMA operation for case 4

Summary

The obtained results from simulation studies shows that the developed EMA concept makes it possible to optimize the usage of the energy generated by WT, which transferred into the costs of energy used in the household. This effect has been obtained by introducing an additional condition (4) in the optimization process, ensuring the best possible usage of the power generated by WT (PWT) to power smart household appliances and improving the comfort of using these appliances by the user. The introduction of specific priority values for individual appliances in the optimization process also affects the achievement of the greatest possible user comfort in using the appliances.

An additional effect of EMA may be the educational function in terms of shaping the user behavior profile in terms of appropriate activation of household appliances when there is the energy generation from RES.

REFERENCES

[1] Powroźnik P., Szcześniak P., Turchan K., Krysik M., Koropiecki I., Piotrowski K., An Elastic Energy Management Algorithm in a Hierarchical Control System with Distributed Control Devices. Energies, 15 (2022), 4750, https://doi.org/10.3390/en15134750
[2] Shewale A., Mokhade A., Funde N., Bokde N.D., A Survey of Efficient Demand-Side Management Techniques for the Residential Appliance Scheduling Problem in Smart Homes, Energies, 15, (2022), 2863, https://doi.org/10.3390/en15082863
[3] Bątkiewicz-Pantuła M., Comparative analysis of production costs of electricity generated from various renewable energy sources, Przegląd Elektrotechniczny, 95 (2019), nr 2, 71-74
[4] Powroźnik P., Szulim R., Miczulski W., Piotrowski, K., Household Energy Management, Appl. Sci., 11 (2021), 1626. https://doi.org/10.3390/app11041626
[5] Powroźnik P., Szcześniak P., Piotrowski K., Elastic Energy Management Algorithm Using IoT Technology for Devices with Smart Appliance Functionality for Applications in Smart-Grid, Energies, 15 (2022), 109, https://doi.org/10.3390/en15010109
[6] Pitsoulisi L., Resende M.G., Greedy randomized adaptive search procedurę, in Handbook of Applied Optimization, Pardalosi P.M., Resende M.G., Oxford University Press: Oxford, UK, (2002), 168–181
[7] https://www.fibaro.com/en/products/wall-plug/
[8] http://www.s-und-w-energie.de
[9] https://www.stromauskunft.de/de/stadt/stromanbieter-injena.html
[10]Erneuerbare-EnergienGesetz,https://www.erneuerbareenergien.de/EE/Redaktion/DE/Standardartikel/FAQ/faq_eeg_2009.html
[11] http://archive.ics.uci.edu/ml/datasets/ElectricityLoadDiagrams20112014

Authors: D.Sc. Wiesław Miczulski, University of Zielona Góra, Institute of Metrology, Electronics and Computer Science, str. Prof. Szafrana 2, 65-965 Zielona Góra, E-mail: W.Miczulski@imei.uz.zgora.pl; M.Sc. Oliwia Pers, ETB Solution GmbH, Dessauer Straße 8, 06-886 Lutherstadt Wittenberg, E-mail: oliwia.pers@etb-solution.com.

Source & Publisher Item Identifier: PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 98 NR 11/2022. doi:10.15199/48.2022.11.28