An experimental approach on linear synthetic inertia

Full text

(1)An experimental approach on linear synthetic inertia Martin Fregelius.

(2) Dissertation presented at Uppsala University to be publicly examined in Häggsalen, Ångströms laboratorium, Uppsala, Thursday, 29 April 2021 at 14:00. The examination will be conducted in English. Faculty examiner: Dr Katarina Yuen. Abstract Fregelius, M. 2021. An experimental approach on linear synthetic inertia. 43 pp. Uppsala: Department of Electrical Engineering, Uppsala University. The interest in renewable energy has significantly increased in the last decades which has led to an increased amount of renewable energy sources in the grid. In the Nordic grid, the major contribution to renewable energy is hydro power and wind power and an increase in the amount of wind power is expected in the future. The increase in wind power and decommissioning of nuclear power is expected to decrease the mechanical inertia in the system which helps to stabilise the electrical grid frequency. The inertia is expected to decrease by a factor of two within 20 years and other solutions for frequency stability must be implemented to assure a stable power system. At Uppsala University several projects are investigating how grid-connected energy storages can increase the frequency stability with a high penetration of intermittent renewable energy sources. In this thesis, a linear synthetic inertia control algorithm is implemented on a national Instruments FPGA for controlling the power flow from a supercapacitor energy storage via a two-level three-phase inverter. The control strategy is evaluated both via simulations and experimental tests in a nano grid. The results of the simulations and experimental work are presented and show that it is possible to calculate the frequency derivative in real time to reduce the frequency ROCOF and nadir. The results of the increased frequency stability are presented. Martin Fregelius, Department of Electrical Engineering, Electricity, Box 534, Uppsala University, SE-751 21 Uppsala, Sweden. © Martin Fregelius 2021 urn:nbn:se:uu:diva-437982 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-437982).

(3) To my family.

(4)

(5) List of papers. This thesis is based on the following papers, which are referred to in the text by their Roman numerals. I. M. Fregelius and U. Lundin, "Hardware implementation of a synthetic inertia system for grid stability," in 2019 8th International Conference on Renewable Energy Research and Applications (ICRERA),pp. 186-190, 2019. II. M. Fregelius and U. Lundin. "Performance evaluation of a super capacitor based synthetic inertia system using frequency locked loop and real time frequency derivative estimation" (Submitted IEEE Open Journal of the Industrial Electronics Society, March 2021). Other contributions by the author of this thesis III. A. Parwal, M. Fregelius, D. C. Silva, T. Potapenko, J. Hjalmarsson, J. Kelly, I. Temiz, J. G. de Oliviera, C. Boström, and M. Leijon, "Virtual synchronous generator based current synchronous detection scheme for a virtual inertia emulation in smartgrids," Energy and Power Engineering, vol. 11, pp. 99-131, March. 2019. IV. J. Abrahamsson, J. Pérez-Loya, M. Fregelius, F. Evestedt, J. Bladh, and U. Lundin, "Magnetic thrust bearing for a 10 MW hydro power generator with a kaplan turbine,"Hydro 2018, 2018. V. U. Lundin, F. Evestedt, J. Abrahamsson, J. Pérez, M. Fregelius, and J. K. Nøland, "Start of a synchronous motor using rotor field polarity inversion and rotor back-emf sensing," in 2020 International Conference on Electrical Machines (ICEM), vol. 1, pp. 338-344, 2020. VI. A. Parwal, M. Fregelius, I. Temiz, M. Göteman, J. G. de Oliveira, C. Boström, and M. Leijon, "Energy management for a grid-connected wave energy park through a hybrid energy storage system," Applied Energy, vol. 231, pp. 399-411, 2018. VII. A. Parwal, M. Fregelius, P. Almeida, O. Svensson, I. Temiz,J. Oliveira, C. Boström, and M. Leijon, "A comparative analysis of linear.

(6) and nonlinear control of wave energy converter for a force control application,"International Marine Energy Journal, vol. 2,pp. 39-50, Jun. 2020 VIII. A. Parwal, M. Fregelius, J. Leijon, M. Chatzigiannakou, O. Svensson,E. Strömstedt, I. Temiz, J. G. de Oliveira, C. Boström, and M. Leijon, "Grid integration and a power quality assessment of a wave energy park," IET Smart Grid, vol. 2, pp. 625-634(9), December 2019. IX. A. Parwal, M. Fregelius, J. Leijon, M. Chatzigiannakou, O. Svensson,I. Temiz, C. Boström, J. G. de Oliveira, and M. Leijon, "Experimental test of grid connected vsc to improve the power quality in a wave power system," in 2018 5th International Conference on Electric Power and Energy Conversion Systems (EPECS), pp. 1-7, 2018. Reprints were made with permission from the publishers..

(7) Contents. 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 1.1 Aim and outline of this thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10. 2. Background. ................................................................................................. 11. 3. Theoretical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Swing equation and linear inertial response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Phase and frequency estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Real-time frequency derivative estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Current control of grid-connected voltage source inverter . . . . . . . . . .. 16 16 17 21 21. 4. Experimental setup for synthetic inertia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Nano grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1 Hybrid Energy storage system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Power electronics and energy storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 23 23 25 25. 5. Synthetic inertia. .......................................................................................... 29. 6. Conclusion. .................................................................................................. 35. 7. Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1 Hybrid energy storage system using bidirectional DC/DC converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Linear Synthetic inertia using the power output of a synchronous generator as reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3 Full scale hybrid energy storage system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 36. 8. Summary of papers. ..................................................................................... 37. 9. Acknowledgement. ...................................................................................... 40. ......................................................................................................... 41. References. 36 36 36.

(8)

(9) 1. Introduction. The energy sector is increasing its amount of renewable energy sources (RES) to meet the increasing energy demand. The effect of increased greenhouse gases [1] has increased the interest in sustainable energy solutions. The main focus on RES has been on hydro power, wind power, and photovoltaic which are among the most developed low pollution production technologies. Other energy sources such as tidal current power, wave power, and marine current technologies are under development and could be future sustainable power sources as well which have gotten a higher interest in recent years.. Figure 1.1. Greenhous gases divided by world regions [1] One common factor for these power sources except hydro power is that the voltage amplitude, current and electrical frequency varies largely over time. To connect energy sources to the grid most commonly power electronics is used to sync to the grid with a maximum power point tracker (MPPT). Due to the non synchronously connected energy sources, the mechanical inertia in the system is reduced. The reduced inertia in the grid will result in a larger 9.

(10) Rate of Change of Frequency (ROCOF) and lower minimum frequencies nadir [2]. Due to the decrease in total power system inertia the risk of getting into frequency regions outside 49.9 Hz and 50.1 Hz is larger, where the most problematic situation is under frequencies. With an increased risk of getting into a low frequency where load shedding and generators will start disconnecting, the interest for synthetic inertia has increased. But due to the problematic estimation of the grid frequency derivative and its long response time the interest for imitating a synchronous generator or virtual synchronous machines has increased. This thesis contributes to the development of a reduced response time when calculating the grid frequency derivative, and using applied power electronics and fast data processing.. 1.1 Aim and outline of this thesis The work presented in this thesis evaluates, through experiments and simulations, synthetic inertia solutions for making it possible to increase the share of renewable energy in the power system. The main focus has been, implementation of control strategies to control grid-connected energy storage via power electronics. The work has contributed so far to 9 papers where 8 are peer-reviewed and published covering different aspects of this field and wave power. The main contribution from the author to the synthetic inertia field comes from paper I and II which will be the main focus in this thesis. The work has mainly been focusing on experiments and laboratory testing of the control strategies. Chapter one will present an introduction to the field of synthetic inertia, chapter two will present the theoretical background of the control strategies and hardware system used in this thesis. Chapter three will present the experimental setup which was designed and built for the nano grid experimental work. Chapter 4 will present the results from the thesis and chapter will conclude the results and chapter 6 will sum up the thesis with the future work.. 10.

(11) 2. Background. The power systems ability to regain a new equilibrium point after being subjected to a disturbance from its original operating state with the entire system intact will be defined as the power system stability in this thesis [3]. The power system stability has been divided into three main sectors, voltage stability, frequency stability, and rotor angle stability. The power system stability will be dived into several sub-sectors as seen in figure 2.1, the main scope of the thesis will be the frequency stability sector.. Figure 2.1. Short term stability and long term stability caused by a disturbance in active power. In the power system a balance between production which is the amount of generated and imported active power and the load amount which will be defined as consumed active power, export active power, and system losses should be equal to keep a steady grid frequency. Therefore the frequency can be used as a measure of the balance between generation and load in the power system which should be 50 Hz in the Nordic grid. Due to an imbalance in active power between generation and load the grid frequency will change. Keeping the frequency within certain limits is vital for the components connected. The grid frequency band limits which define the normal operation and disturbed operation for the European countries are defined by the European Network of 11.

(12) Transmission System Operators (ENTSO-E) which is consisting of 43 Transmission system operators from 36 countries. The grid codes define which action should be taken when the frequency deviates from the base frequency [4], [5]. These grid codes are implemented in each country and designed to reduce the risk of system collapse or damage on components. An example could be disconnection of a large load which could result in an over frequency where generators need to disconnect or a generation loss from either a large power plant or HVDC-line which would lead to under frequency where load shedding could be necessary or in the worst case lead to a total blackout. The power system frequency stability will be defined as the ability to keep the frequency within limits after a large disturbance either by a loss in generation or loss in load. The frequency stability will be divided into short-term and long-term stability. Were short term stability is defined by the total inertia in the system and lasts within a couple of seconds, and the long-term stability is defined to several minutes, seen in figure 2.2.. Figure 2.2. Short term stability and long term stability caused by a disturbance in active power. The frequency stability is dependent on three factors, inertia, available Frequency Containment Reserve (FCR), and power imbalance. A large power imbalance or small system inertia and frequency reserves will lead to a larger and faster frequency change. In a power system, the largest imbalance is the dimension factor, in the Nordic grid, is 1450 MW due to the nuclear power plant in Oskarshamn which will set the amount of reserves needed. However recent blackouts in Great Britain have shown that large wind farms can also contribute to a large power deviation due to bad weather situations [6] and cause blackouts if not taken into consideration as a dimension factor. 12.

(13) During a power deviation between generation and load the behavior of the frequency during the first moment can be described by the inertia in the system. During the first moment, the power imbalance will be compensated by the release of the kinetic energy stored in the rotating parts of the synchronous generators or synchronous compensators connected to the grid. The amount of energy that is taken from the kinetic energy will decrease the rotation speed of the generators which will result in a decrease in electrical frequency. Usually over frequency is not a big problem since the generated power can be decreased, under frequencies are harder to tackle since the amount of power generated is too small and need to be increased before the frequency falls too much and load shedding and generation disconnection start happening. To mitigate the frequency change, in the Nordic grid Frequency Containment Reserve (FCR)-Normal and FCR-Disturbed are used as the frequency reserve which usually starts acting after several seconds, the activation times of the reserves are shown in figure 2.3. The FCR, aFRR, mFRR and RR are usually supplied by hydro power. The FCR-N is responsible for keeping the frequency within its normal band which is defined as 50.0(1) Hz which is most commonly done with a PI regulator with droop and is fully activated at 49.9 Hz. The FCR-D starts acting at 49.9 Hz and is fully activated at 49.5 Hz.. Figure 2.3. Short activation time for the frequency reserves in the Nordic grid. Starting with the fastest to left and and decreasing in activation time to the right.. In 2020 a new frequency reserve has been added, Fast Frequency response (FFR) which acts within one second depending on the type. The FFR can be supplied either by a fast energy source such as batteries or a fast load which can decrease its power consumption. The FFR must have a duration time of a minimum of 30 s [2]. The main scope of the thesis will be the inertia response of the power system where decreasing mechanical inertia is a concern in most power systems today. The help from inertia to suppress fast and large frequency fluctuations has decreased significantly in the last decade due to the increased amount of RES. Svenska Krafnät is expecting a large decrease in system inertia, 300 GWs to 150 GWs between year 2020 and 2040 [7]. The decrease of inertia in the Nordic grid comes mainly from the decommissioning of nuclear power and an increase of wind power. The decommissioning of the synchronous generators and replacing them with power electronics connected wind turbines, most commonly doubly-fed induction generators which 13.

(14) are not electrically coupled to the grid frequency, decreases the total system inertia since they do not contribute a large amount of inertial response to grid frequency deviations. However the active power setpoint for the power electronics can be changed based on the frequency derivative which has been evaluated in [8],[9] and [10]. One of the drawbacks of using the mechanical inertia in the wind turbine and coupling to the grid frequency is the power reduction after the frequency disturbance needed to restore the optimal blade speed. The reduced power output after the disturbance could cause a second frequency dip if not handled, an optimal use of the kinetic energy in variable wind turbines has been discussed in [11]. Hydro-Québec in Canada has already defined the need of inertia in their grid codes and have had long-term tests with inertia response from wind power [12],[13]. In the inertia emulating wind turbines, there is a predefined state when the turbine shall add extra power to the grid, mainly when the frequency drops outside of the normal band which will help the system to decrease the nadir and ROCOF. If the grid inertia is reduced significantly the frequency quality will affect the operational pattern for FCR which may lead to an increase in regulation from the generation units. The increase in regulation will lead to an increase in wear and tear of the mechanical parts, which will result in an increased maintenance cost [14]. A wide range of solutions for decreasing grid inertia has been investigated by using either synchronous compensator or grid-connected energy storage via power electronics. Synchronous compensator have been used for a long time in the power grid. Where the synchronous compensator only has a spinning rotor without any connected turbine. The magnetizing current in the rotor can be controlled to handle reactive power compensation. The synchronous compensator is also increasing the short circuit capacities which help to stabilize week grid conditions. More recently the contributions to low inertia situations have started to be evaluated [15], where it has been shown that in grids with high penetration of RES the inertia from the synchronous compensator helps to increase the frequency stability, both long term, and short term. In recent years power electronics have increased in interest for fast frequency control and its ability to connect energy storages to the grid. Since the computational power no longer is a limiting factor when using Field Programmable Gate arrays (FPGAs) or Digital Signal Processors (DSP) more advanced control schemes can be implemented. Most commonly know are Virtual Synchronous Machine (VSM) or (VISMA), Virtual Synchronous Generator (VSG), where the aim is to imitate the behavior of a synchronous generator with a RES by adding the inertia response, damping and voltage control which has been discussed in [16],[17] and [18]. Inertia emulation using HVDC-lines shunt capacitors as short-term energy storage and evaluating the correlation between the dc-bus voltage and the frequency derivative has been discussed in [19]. It was shown that the correlation between the DC-link voltage and frequency derivative could be used to avoid 14.

(15) the need to calculate the frequency derivative. Due to the inertia is dependent on the frequency derivative, and the frequency estimation contains noise the unwanted behavior of amplified noise while calculating the frequency derivative will be large. Where the most common way to calculate the frequency in grid-tied inverters, is by using a phase-locked loop (PLL), which estimates the phase of the grid voltages. An extra feature from the PLL is the estimation of grid frequency also, but since it’s optimized for reducing the error on the phase the frequency will contain high-frequency noise. In [20] an analysis of the possibility to use an energy storage with an inverter and estimating the derivative in real-time the result was that the Signal to Noise Ratio (SNR) was low, by low pass filtering the frequency over one second it was possible to estimate the frequency derivative. The filter delay caused thereby a significant phase delay of the system which made it impractical.. 15.

(16) 3. Theoretical background. In this chapter, the theoretical background is presented of the system. This chapter starts with explaining the need of inertia in section 3.1 and how mechanical inertia can be compared with synthetic inertia. In section 3.2 the frequency and phase estimations used in the project are explained. To calculate the real-time frequency derivative a special Finite Impulse Response (FIR) filter has been used which is explained in section 3.3. The estimated grid frequency derivative is acting as a reference for the amount of power supplied to the grid, where the power is converted to a reference grid current and realized by the fast current controller explained in section 3.4.. 3.1 Swing equation and linear inertial response To quantify the need of synthetic inertia we need to understand the dynamics of the synchronous machine which connects to the system frequency. Due to the rotating masses in the synchronous generators the dynamic of the system frequency can be described by the sum of all synchronous generators where one generator is described by the swing equation 3.1. Ji. dωm,i = Tm,i − Te,i dt. (3.1). Where Ji is the total moment of inertia of the generator in kg · m2 , Tmi is the mechanical torque in Nm and ωi is the mechanical angular speed of the rotor. By multiplying equation 3.1 with ωmi the following equation is obtained ωmi Ji. dωmi = Pmi − Pei dt. (3.2). where Pei is the electrical power fed to the grid and Pmi is the mechanical power from the turbine. The mechanical speed is connected to the electrical frequency via fel = 2p fmech where p is the nuber of poles in the machine and fmech is the mechanical angular speed of the rotor. We then define the inertia constant as H=. stored kinetic energy at synchronous speed generator rating. (3.3). Jωm2 2Sn. (3.4). H= 16.

(17) Where different types of generators have different inertia constants, usually varying between 1-10 seconds. Where hydro power usually has 2−4s, nuclear power 4 − 10s [21]. Inserting the inertia constant into the swing equations results in 2Hi Sni = Pmi − Pei 2 ωmsi. ωmi. (3.5). Rewriting equation 3.5 and assuming the rotor speed does not differ considerably from the synchronous speed the equation can be rewritten as 2Hi Sni d fi = Pmi − Pei fs dt. (3.6). Where the change in system frequency will be dependent of the total system inertia where the total system inertia is described by equation 3.7. H=. ∑ni=1 Hi Sni ∑ni=1 Sni. (3.7). By rewriting equation 3.6 and inserting the total system inertia the dynamics of the power system can be described by 2. HSn dΔ f = ΔP fs dt. (3.8). where ΔP is the power output of the synthetic inertia system and Δ f is the difference in frequency from 50 Hz. Synthetic inertial response Since the synthetic inertia has no rotating parts the inertia response will be given by the power output at a specific frequency derivative. Rewriting equation 3.2 to 3.9 the inertia response can be compared with a synchronous generator. PSI = KSI where. KSI = J. d fel , dt. 2π nP. (3.9). 2 fel .. (3.10). 3.2 Phase and frequency estimation To grid connect an energy storage, the voltage phase angle, voltage amplitude, grid frequency, and phase order must be estimated to sync the inverter to the 17.

(18) grid. To estimate the phase of the three-phase grid voltages an optimized Double Second-Order Generalised Integrator Phase Locked Loop (DSOGI-PLL) was used which is described in the first paragraph below. For estimating the grid frequency an optimized DSOGI Frequency Locked Loop (DSOGI-FLL) was used which is described in detail in the second paragraph below. Phase locked loop The most common way to estimate the phase in grid-tied inverters is to use a synchronous reference frame phase-locked loop (SRF-PLL)[22],[23]. It has a simple digital implementation structure which makes it easy to tune the parameters of the PLL [24]. The basic function of the PLL is to estimate the voltage phase angle from the angular frequency of the grid voltages. The grid voltages for a symmetric three-phase system are electrically phase shifted by 120° from each other. The voltages are converted from Va Vb Vc to Vd Vq via a two step transformation. First via a Clark transformation equation3.11 to Vα Vβ and then to Vd Vq with equation 3.12 a park transformation. ⎡ ⎤ Va 1 √ −1/2 −1/2 Vα √ ⎣Vb ⎦ (3.11) = Vβ 0 3/2 − 3/2 Vc Vd Cosθgrid −Sinθgrid Vα = (3.12) Vq Sinθgrid Cosθgrid Vβ Where the α and β are in the stationary two-dimensional reference frame and the dq are in the rotating two-dimensional reference frame, where the reference frame is rotating with the grid angular frequency. When substituting Va and V b,V c with its respective phase shifted values we can rewrite equation 3.11 as follows Vα peak Cosθgrid =V (3.13) Vβ −Sinθgrid and then applying the Park transformation and simplifying we get equation 3.14. Vd = V peakCos(θgrid − θ ) Vq = V peak Sin(θgrid − θ ). (3.14). To be able to achieve synchronisation to the grid, the estimated θ must be equal to θgrid which results in equation 3.14 will get Vd = V peak , Vq = 0. By assuming that the error between θ and θgrid will be small we can simplify sin(θgrid − θ ) = θgrid − θ . When designing the proper controller the error can be kept small and thereby estimate the grid phase. But due to grid faults and nonperfect sinusoidal grid voltages, the SRF-PLL will be distorted by negative sequence components and grid harmonics. This 18.

(19) will lead to an error in the estimation of the grid phase. To minimize the oscillation and ripple in the phase estimation a Double Second-Order Generalised Integrator (DSOGI) pre-filter can be used. The DSOGI was used for filtering and obtaining the 90° shifted versions from the Vα and Vβ signals. The Vα ,Vβ signals were the inputs for the DSOGI filter and the outputs were used as inputs for the Positive-negative-Sequence calculator (PNSC). The positive voltage sequence is then used as input for the SRF-PLL. The transfer function of the SOGI is described in equation 3.15 and show in figure 3.1. SOGI(S) = . . .

(20) . . .

(21) . ω s s2 + ω 2. . (3.15). . . . . . . . . . . . . . . . .

(22) . . . . . . . . Figure 3.1. Shows the block diagram of the Double Second Order General Integral Phase Locked Loop II. Were the top left part shows the Clark transformation and the DSOGI prefilter. The top right part shows the synchronous reference frame PLL.. The transfer functions of the in quadrature signals v and qv are shown in equation 3.16 and 3.17 where the behaviour of the band-pass filter D(s) and low-pass filter Q(s) can be seen. D(s) =. v kω s (s) = v s2 + kω s + ω 2. (3.16). qv kω 2 (3.17) (s) = v s2 + kω s + ω 2 The settling time of the QSG-SOGI is defined by K in equation 3.18 10 (3.18) ts = (SOGI) kω Q(s) =. 19.

(23) Frequency locked loop Instead of using a PLL for estimating the frequency a similar way is to use a Frequency Locked Loop (FLL) which is useful in many applications [25]. The PLL is optimized for tracking the voltage phase of a sinusoidal signal and by doing it the frequency can easily be estimated, but it will be sensitive to voltage unbalances and have a high noise content. Instead, an FLL can be used which only locks on the frequency of an input signal and minimizes the frequency estimation error [26],[27]. The FLL is shown in figure 3.2. . .

(24) . . . .

(25) . . . . .

(26) . . . . . . . . . . . . . Figure 3.2. Shows the block diagram of the Frequency Locked loop II. Where the top part shows the Clark transformation, DSOGI prefilter, and the positive sequence estimator. The middle part shows the block diagram of the DSOGI and the bottom part shows the frequency estimator.. The DSOGI filter was used as a prefilter to estimate the frequency accurately during grid faults or harmonics. The DSOGI-FLL uses the positive sequence Vα Vβ together with the product off qv and the error signal ε. Were the transfer function from V to ε is given by equation 3.19. E(s) =. s2 + ω 2 s2 + kω s + ω 2. (3.19). Due to the nonlinear behavior of the FLL linear control analysis can not be applied directly. By linearizing the FLL through normalizing with the factor kω Γ it will become independent of grid variables and the SOGI gain. The setV2 20.

(27) tling time can then be described by equation 3.20 and will only be dependent on Γ ts (FLL) ≈. 5 Γ. (3.20). 3.3 Real-time frequency derivative estimation Calculating derivatives on a continuously noisy signal is mostly avoided due to its amplifying behavior. To be able to calculate the frequency derivative a Finite Impulse Response (FIR) Filter was used. The FIR filter can be described in the frequency domain with the convolution theorem 3.21 F(x ∗ h) = F(x) · F(h). (3.21). where F(x) is the sampled signal and F(h) is the filter function. The impulse response can be described by N. h[n] = ∑ an · δ [n − i],. (3.22). i=0. where an are the filter coefficients and δ is the sampled signal. The filter coefficients can be precalculated and stored in a Look-Up Table (LUT) for increased processing speed. The coefficients used in this case are based on a Savitzgy-Golay (SG) filter and is discussed in [28],[29],[30]. The SG filter is used for running a real-time least-square polynomial fitting around one point. The filter will act as a smoothing filter but preserving the characteristics of the original signal. By choosing the order of the fitted signal and the window width a certain smoothing can be accomplished. The phase shift will be linear with the time delay of the window width. By choosing the correct order of the polynomial, the derivative of the polynomial can be calculated which will result in that the convoluted signal will be the estimated derivative.. 3.4 Current control of grid-connected voltage source inverter The most common way to connect grid-following Voltage Source Converters (VSC) is by using a Voltage Oriented Control (VOC) strategy. The reference voltage is calculated from the grid currents in the rotating dq reference frame [31]. Were the direct axis current controls the active power fed to the grid, and the quadrature axis controls the reactive power fed to the grid. The performance of the VSC will mainly depend on the PWM technique used to calculate the references. In this case, a sinusoidal PWM technique is used. To 21.

(28) control the dq grid currents two PI regulators are used, which is possible due to transformed grid currents will be DC values. The PI regulator will bring the error between the reference currents id ∗ , iq ∗ and the real inverter currents id , iq to zero. The output voltage of the inverter can be described mathematically by equation 3.23 and 3.24 md = −L. did + Rid ωLiq + vd dt. (3.23). diq (3.24) + Riq ωLid + vq dt Where md and mq are the output dq components that are fed into the reverse transformation. It can be seen that the output of the d and q currents are coupled, to be able to control the independently a decoupling term is added which is dependent on the total inductance in the filter. A feed-forward for the grid voltage is added to improve performance. mq = −L. 22.

(29) 4. Experimental setup for synthetic inertia. For the purpose of testing and validating new ideas, an experimental generator has been designed and built during the last 10 years by the hydro power group at Uppsala University. The setup has several ongoing experiments due to its flexibility and will be explained in detail in section 4.1. During the past 3 years, an experimental hybrid energy-storage has been designed and built for testing different grid ancillary services in the nano grid which is described in section 4.2.. 4.1 Nano grid To be able to test and simulate a nano grid the system consists of three main parts, a constant resistive load, a variable resistive load, and a synchronous generator. The parts are described in more detail below and a block diagram of the system is shown in figure 4.1. The system can operate in both island and grid-connected mode depending on the tests being performed. The 400 VAC point of common coupling (PCC) is equipped with a synchronization system that can connect the system to the main grid. Prime Mover An ABB frequency converter controls an asynchronous machine rated 55kW. The motor shaft is connected to the synchronous generator shaft via a 3 : 1 90° gearbox. The motor acts as the prime mover for the generator and provides the driving torque. The converter is provided with a torque reference from the main controller. Synchronous generator The generator is based on a synchronous salient pole machine seen in figure 4.3. The rotor shaft is connected to the asynchronous motor via a gearbox where the motor provides the driving torque. The axial force of the rotor and shaft are unloaded via a permanent magnet bearing and an electromagnet. The radial position of the shaft is held in place by two mechanical bearings. For the magnetization of the generator two systems can be used, the first on the top of the generator shaft, which is a rotating magnetization system based on rotating IGBT power electronics. The second one is using the slip rings for magnetization with an external DC power supply. The terminals of the 23.

(30) .

(31) . . . . . . . Figure 4.1. An block diagram overview of the nano grid setup, consisting of a synchronous generator, hybrid energy storage, a constant resistive load and a variable resistive load.. generator are connected to a step up transformer to connect to the 400 VAC bus. The parameters for the generator can be seen in table where the inertia is the total inertia from the rotating prats in the setup. 4.1 Table 4.1. Synchronous generator parameter specification Component Rated power Rated voltage Nominal speed Nominal frequency Number of poles Inertia. 24. value 70 kVA 156 V 500 RPM 50 Hz 12 ≈ 100kgm2.

(32) Figure 4.2. National Instruments cRIO Control system for the prime mover FCR and AVR.. Load The load is based on two three-phase resistors. The constant load has a resistance of 9.6 Ω and can consume 50 kW. The low power 6 kW load is controlled by a variable transformer which changes the output voltage to the Wye connected three-phase resistor. The turn transformer is changed by a step motor which can increase the voltage from 0 % to 100 %, where a predefined load variation can be simulated.. 4.1.1 Hybrid Energy storage system. 4.2 Power electronics and energy storage The hybrid energy storage, seen in figure 4.5, consists of a Lithium-Ion battery pack and a Maxwell technologies supercapacitor, connected to the grid via a three-phase Semikron inverter with an LC filter. The specifications of the parts can be found in table 4.2. The power system is grid-connected via an ΔY step-up transformer to 400 VLL which is designed to be able to output 15 kVA to the grid. A 400 V system base voltage was chosen due to the nano grid, which made it possible to use stan25.

(33) Figure 4.3. Shows the generator setup used for creating the 400 V nano grid !" " .

(34) . #.

(35) . %!

(36) . . . . #$. Figure 4.4. Shows an overview of the hybrid energy storage system, with two energy storage systems and a two level inverter with its control system.. dard components. The transformer inductance together with the LC filter will create the LCL filter which reduces the voltage THD within the grid limits. An 68 F 125 V supercapacitor was chosen as the power storage which can handle 1 000 000 charge and discharge cycles. Lithium-Ion batteries were chosen for long-term energy storage. Hybrid energy storage control system The controller used for the energy system is based on a National Instruments Single-Board RIO (sbRIO) 9607 with a general-purpose inverter controller 26.

(37) Table 4.2. Hardware system parameters for the hybrid energy storage Component Battery voltage Battery capacity Super-capacitor voltage Super-capacitor capacitance Inverter rating Filter inductance Filter capacitance Transformer rating Transformer Voltage. value 180V 7.2 kWh 125 V 68 F 1200 V, 250 A 95 μH 50 μF 15 kVA 400 V/65 V. Figure 4.5. Overview of the hybrid energy storage power cabin. On the left side the three-phase Semikron inverter connected to an LC filter and transformer, and on the right side the Maxwell super-capacitor and Lithium-Ion energy storage.. card (GPIC) 9684 seen in figure 4.6. The GPIC card is connected to an interface card which collects all signals and converts them to the correct voltage levels. The controller contains a Field Programmable Gate Array (FPGA) which has a base clock frequency of 40MHz and a real-time system based on UNIX, which handles data communication and data transfer. 27.

(38) Figure 4.6. Shows the control system based on National Instruments Single board RIO with a general-purpose inverter controller card (GPIC) 9684.. 28.

(39) 5. Synthetic inertia. The results will be divided and presented in the order of the papers in the paragraphs below. paper I A nano grid was used for testing the SI system by step-wise increasing the total inertia in the system containing both the mechanical inertia from the SG and the synthetic inertia from the SI system seen in figure 5.1.. ܲ௅. ͳ ‫ ݏܯ‬൅ ‫ܦ‬. SI. FCR. Controlable Load. ܲி஼ோ. Δf. ܲௌூ ΔP. +. Figure 5.1. Shows the nano grid used with the synchronous generator, controllable load and the SI system, all connected to a common 400 V AC bus. In this case the synchronous generator will provide with both mechanical inertia, damping and frequency control, FCR, paper I.. The grid frequency was estimated in real-time and the derivative was calculated in real-time using an SG filter. A predefined load change seen in the bottom part of figure 5.2 was used to create a frequency change in the nano grid. The load change was ±1250 W with a base load of 19 kW. The top part of figure 5.2 shows the frequency response of the nano grid for three different amounts of SI inertia. The frequency variations are large compared to the Nordic grid in this case but this is not a problem in the nano grid and is sufficient for a proof of concept. 29.

(40) Frequency [Hz]. 50.5 Without SI SI 3500 SI 6500. 50. 49.5. 49 0. 10. 20. 30. 40. 50. 60. 40. 50. 60. Time[s]. Variable load [W]. 1000 500 0 -500 -1000 0. 10. 20. 30. Time[s]. Figure 5.2. The bottom graph shows the load change over time which was used in the test cases. The top graph shows three different total inertia cases. The blue is without any extra inertia from the SI system, the red is with a 64% increase in inertia and the yellow is with and 119% increase of total inertia. The base inertia is 100 kgm2 , paper I.. It can be seen that when the inertia in the system is increased it acts as a low pass filter and reduces the minimum frequency nadir and reduces the maximum frequency ROCOF. The data is shown in table 5.1. Table 5.1. Frequency data with no added synthetic inertia and with synthetic inertia added, paper I Case. Δ fmax [Hz]. df dt max [Hz/s]. Nadir [Hz]. Without SI. 0.83. 0.4431. 48.94. With SI 64%. 0.8. 0.3184. 48.97. With SI 119%. 0.47. 0.2605. 49.3. paper II An FFT analysis was done of the grid voltages, estimated frequency, and frequency derivative estimate, to be able to adapt the filter parameters of the DSOGI and low pass filter. The filter parameters were adapted to filter out higher harmonics and minimizing the impact on lower frequencies which were the wanted signals. In the top part of figure 5.3, it is shown that the voltage contains several harmonics. In the two middle graphs, the estimated grid frequency FFT is shown, it can be seen that some harmonics are present and at lower frequencies the is noise. The noise in the lower frequencies is getting amplified when calculating the frequency derivative. This can be seen in the bottom graph. 30.

(41) Voltage Spectrum. |V(f)|. 2 1 0 0. 50. 100. 150. 200. 250. 300. 350. 400. 450. 500. 350. 400. 450. 500. 350. 400. 450. 500. 350. 400. 450. 500. Frequency spectrum. |f(f)|. 0.5 0 0. 50. 100. 150. 200. 250. 300. Filtered Frequency spectrum. |f(f)|. 0.05. |df/dt(f)|. 0 0. 50 10. 100. 150. -4. 2 1 0 0. 200. 250. 300. Dervivative spectrum. 50. 100. 150. 200. 250. 300. Frequency (Hz). Figure 5.3. Shows the Fast Fourier Transform of the grid voltages in the top, estimated frequency, filtered frequency estimation which in this case is the same as the estimated frequency and the estimated grid frequency derivative in the bottom, paper II.. To filter the noise a low pass filter is introduced which is filtering the grid frequency estimate. A third order low pass filter with a cutoff frequency at 15 Hz is implemented. It can be seen in figure 5.4 that the estimated frequency derivative has reduced the amount of noise in the lower frequency region and all higher frequencies are damped.. Voltage Spectrum. |V(f)|. 2 1 0 0. 50. 100. 150. 200. 250. 300. 350. 400. 450. 500. 350. 400. 450. 500. 350. 400. 450. 500. 350. 400. 450. 500. Frequency spectrum. |f(f)|. 0.5 0 0. 50. 100. 150. 200. 250. 300. Filtered Frequency spectrum. |f(f)|. 0.05. |df/dt(f)|. 0 0 2 1 0 0. 50 10. 100. 150. -4. 200. 250. 300. Dervivative spectrum. 50. 100. 150. 200. 250. 300. Frequency (Hz). Figure 5.4. Shows the Fast Fourier Transform of the grid voltages in the top, estimated frequency, low pass filtered frequency estimation and the estimated grid frequency derivative in the bottom, paper II.. 31.

(42) In order to evaluate the performance of the system, a three-phase voltage was simulated in National Instruments Labview with 120° phase shift between the voltages and an RMS voltage of 230 VLN . The frequency of the three-phase voltages could be controlled to make it possible to create a stepchange in frequency or add an overlaid signal with a specific amplitude A f˜ and frequency f˜, equation 5.1..

(43) (5.1) Va = Va0 sin 2π 50 + A f˜ sin(2π f˜) In figure 5.5 a −0.4 Hz step change from 50 Hz is introduced and the frequency estimation, low pass filtered frequency are shown. In the bottom graph, the estimated grid frequency derivative is shown.. Figure 5.5. A grid frequency step of 0.4 Hz is introduced. In the top graph the frequency step and the estimated frequency is show. In the bottom graph the estimated frequency derivative is show with varying sample frequencies between 10 kHz and 1.6 kHz, paper II.. The phase delay from the frequency estimation can be seen which is approximately 40 ms, the low pass filter introduces a significant phase delay with a total time delay of 0.2 s. The bottom graph shows the phase delay introduced by the grid frequency derivative estimation done by the SG filter. It can be seen that with a fixed window width the phase delay introduced will be dependent on the sample time reaching from 0.35 s to 1.2 s. It can be seen in the estimated frequency derivative that the amplitude decreases due to the total time span the estimation is done over. The system was grid-connected to the main grid with the SI controller activate. At the top of figure 5.6 the estimated grid frequency of the main grid is shown. It can be seen that the changes in frequency are slow and small for the Nordic grid. In the middle graph, the estimated grid frequency derivative 32.

(44) is shown and is varying between ±5 mHz/s. In the bottom graph, the active output power of the SI system is shown. It can be seen that the power follows the frequency derivative estimation with an amplification.. Frequency Filtered frequency. [Hz]. 50.02 50 49.98 49.96 49.94 20. 40. 60. 80. 100. 120. 140. 160. Derivative. [mHz/s]. 5. Labview derivative. 0. -5 20. 40. 60. 80. [W]. 100. 120. 140. 160. Power. 1000. Poweroutput. 0. -1000 20. 40. 60. 80. 100. 120. 140. 160. Time [s]. Figure 5.6. Shows the system response while grid connected to the main grid. In the top graph the estimated grid frequency is shown, in the middle the estimated grid frequency derivative is shown and in the bottom graph the active output-power is shown, paperII.. The system was simulated in Matlab Simulink using the same sampled voltage signal used in the real test to verify the results of the individual filters and the total SI system. The three-phase grid voltages were sampled with 10 kHz, also the estimated values were sampled with the same frequency and saved for comparison with the ideal filters in Matlab Simulink. The Simulink control scheme is shown in figure 5.7 with the top part showing the DSOGI-PLL, the middle part shows the grid FLL together with the SG filter. The bottom part of the figure shows the VOC which is creating the gate pulses for the inverter. The simulation is using the same filter parameters as in the real system and the estimated values had a high correlation with the real-time estimated values. Some smaller time delays were observed in the real-time estimations of the frequency and derivative which were expected due to processing time. 33.

(45) Figure 5.7. Shows the control loops for the Matlab Simulink inverter implementation. The top part shows the PLL which used for phase angle estimation, the middle part shows the frequency estimation and the bottom part shows the voltage oriented control for the inverter.. For the simulation, a fixed DC supply was used since only the filter parameters were the main scope of the simulation. In figure 5.8 the topology used for simulation is shown. From the left, the constant DC source is shown, connected to a 2 level voltage source converter with an LC filter on the output. Due to the low voltage of the DC source, a Δ/Y step-up transformer was used from 65 V to 400 V to increase the voltage and to reduce the third harmonic into the grid. A strong grid was connected on the high voltage side of the transformer.. Figure 5.8. Shows the Matlab simulink simulation of the 2 level voltage source converter, LC filter and Δ/Y transformer connected to a strong grid.. 34.

(46) 6. Conclusion. A 15 kVA, 400 V synthetic inertia energy storage system was designed, built and tested in a nano grid setup. An estimation of the grid frequency in realtime with small phase delay and high accuracy was implemented in a National Instruments sbRIO FPGA controller using a frequency locked loop. The frequency signal was used for estimating the grid frequency derivative in realtime using an SG filter to create a linear synthetic inertia controller. It was shown that the frequency nadir could be minimized by increasing the total system inertia containing both mechanical inertia and synthetic inertia. The ROCOF was significantly reduced by the synthetic inertia system in the nano grid. An evaluation of the filter parameters was done using the main grid where the phase delay introduced by the FLL could be reduced but with an increased amount of noise in the estimated grid frequency. A settling time of 50 ms was chosen to be fast enough but with enough reduction of the noise in the frequency estimation. It was shown that the window width and sample frequency of the FIR filter affected the phase significantly and respect to the amount of inertia in the power system needed to be taken. The smaller the amount of inertia the faster the frequency changes are expected which will need a smaller window size and higher sampling frequency to reduce the phase delay caused by the FIR filter. The 90° phase shift introduced by the frequency derivative estimation must also be taken into consideration since this will result in that the phase margin will be reduced significantly.. 35.

(47) 7. Future work. 7.1 Hybrid energy storage system using bidirectional DC/DC converters The test setup will be continued by expanding the synthetic inertia system capability with a bidirectional DC/DC converter. The DC/DC converter will make it possible to control the power flow from each energy storage which in this case will be from the super capacitor and Li-ion batteries. With a hybrid energy storage system the effects of fast frequency control and linear synthetic inertia can be compared in the nanogrid.. 7.2 Linear Synthetic inertia using the power output of a synchronous generator as reference Instead of using the frequency of a PLL or FLL the power output of a synchronous generator could be measured. This would reduce the phase delay caused by the PLL or FLL and derivative estimation. This can be referred to equation 3.9 where the short time power difference from the generator could be amplified. A test in the nano grid could be done to test the difference between a mechanical amplified power difference from the generator and the inertia response from the SG filter.. 7.3 Full scale hybrid energy storage system The concept of using a energy storage combined with a hydro power station for faster frequency control is going to be implemented in full scale at a power station. The station has a maximum power output of 24 MW and will be combined with a energy storage to reduce the wear and tear of the mechanical parts and to fulfil the frequency regulation requirements.. 36.

(48) 8. Summary of papers. This chapter summarises the papers and describes the author contribution to each paper.. Paper I Hardware implementation of a synthetic inertia system for grid stability This paper describes the the hardware system built using super capacitors and a 2-level voltage source converter, and the implementation of a real time frequency derivative estimating algorithm. Data from an emulated load change in the nano grid using synthetic inertia together with the mechanical inertia from the synchronous generator are presented. The author did designed, build and did the experimental work, carried out the simulations, validated the data and wrote the paper. The load profile was created by Elin Dahlborg and should replicate fast and slow load changes in the grid as well as some step changes in the grid load.. The contribution was presented by the author at 8th International Conference on Renewable Energy Research and Applications (ICRERA). Paper II Performance evaluation of a super capacitor based synthetic inertia system using frequency locked loop and real time frequency derivative estimation This paper evaluates the performance of the synthetic inertia system and how the filter parameters affect the total phase delay from a change in grid frequency to change in active power output. The main parameters that which were optimised, was the frequency locked loop settling time, Savitzgy-Golay filter window width and sampling frequency. The author did perform all experimental work, validated the data and wrote the paper. Paper submitted to IEEE Open Journal of the Industrial Electronics Society. Paper III Virtual synchronous generator based current synchronous detection scheme for a virtual inertia emulation in smart grids 37.

(49) The paper presents a inertia control using the energy from the DC link capacitors which handles the fast frequency transients. A virtual synchronous generator scheme with droop control for the frequency deviation and voltage deviation at the point of common coupling was used. The author was involved in discussions. The paper is published in an issue of Clean Energy of Energy and Power Engineering, 2019.. Paper IV Magnetic thrust bearing for a 10 MW hydro power generator with a kaplan turbine In this paper the design, experimental work and commissioning och a electromagnetic thrust bearing for 10 MW turbine in Porjus was reported. The author was involved in discussions, experimental work and commissioning. Presented by Johan Abrahamsson at Hydro 2018. Paper V Start of a synchronous motor using rotor field polarity inversion and rotor back-emf sensing The author contributed with discussions during the project and in the reviewing of the paper.. Paper VI Energy management for a grid-connected wave energy park through a hybrid energy storage system This paper presents and energy management system, combining battery and super capacitors. The energy management system uses a fully active topology which optimises the power flows from each energy storage and increases lifetime from the batteries. The author was involved in discussions. The paper is published in Applied Energy, 2018.. Paper VII A comparative analysis of linear and nonlinear control of wave energy converter for a force control application 38.

(50) This paper presents a control strategy for a permanent magnet linear generator by controlling the the stator currents. An evaluation between a linear PI controller and a nonlinear controller i presented where a neural model is implemented. The author contributed to discussions. The paper is accepted in International Marine Energy Journal.. Paper VIII Grid integration and a power quality assessment of a wave energy park This paper presents the results from experimental work of the grid connection system used at the test site in Lysekil Sweden. The system was based on a 2-level voltage source converter connected to the grid via a LC filter and a tap-transformer. The paper presents the power quality at the point of common coupling. The author did contribute to designing, building, programming and testing the system. The paper is published in IET Smart Grids 2019. Paper IX Experimental test of grid connected VSC to improve the power quality in a wave power system This paper presents the details of the grid connecting system design, control and measuring system. An analysis of the Total harmonic distortion at the PCC was done to verify the fulfilment of the grid code. The author did contribute to designing, building, programming and testing the system. The paper was presented by A. Parwal at 5th International Conference on Electric Power and Energy Conversion System (EPECS).. 39.

(51) 9. Acknowledgement. The research presented in this thesis was carried out as a part of Swedish Hydropower Centre - SVC. SVC has been established by the Swedish Energy Agency, Energiforsk and Svenska Kraftnät together with LuleåUniversity of Technology, KTH Royal Institute of Technology, Chalmers University of Technology, Faculty of Engineering, Lund University and Uppsala University. Participating companies and industry associations are: AFRY, Andritz Hydro, Boliden, Fortum Sverige, Holmen Energi, Jämtkraft, Jönköping Energi, Karlstads Energi, LKAB, Mälarenergi, Norconsult, Rainpower, SkellefteåKraft, Sollefteåforsens, Statkraft Sverige, Sweco Energuide, Sweco Infrastructure, Tekniska verken i Linköping, Uniper, Vattenfall R&D, Vattenfall Vattenkraft, Voith Hydro, WSP Sverige and Zinkgruvan. I would like to thank my supervisor Urban Lundin for always taking the time to have interesting discussions and letting me explore the path of a PhD. It has been a great time with many challenges, some stressful time and allot of lessons learned. No matter which strange idea or question you have taken the time do discuss it and help me navigate to a result. Your guidance and help has been invaluable. I would also like to hank my assistant supervisor Johan Abrahamsson who has been a great help with valuable inputs and discussions both about work and other strange and interesting subjects. Also a big thank you to the Hydro power group for interesting discussions and all the travels to Pörjus. To my family, thank you for all the support.. 40.

(52) References. [1] “Cumulative COâ emissions by world region.” (2021, March 13). [2] “Future System Inertia2.” (2021, March 13). [3] N. Hatziargyriou, J. V. Milanovic, C. Rahmann, V. Ajjarapu, C. Canizares, I. Erlich, D. Hill, I. Hiskens, I. Kamwa, B. Pal, P. Pourbeik, J. J. Sanchez-Gasca, A. M. Stankovic, T. Van Cutsem, V. Vittal, and C. Vournas, “Definition and classification of power system stability revisited extended,” IEEE Transactions on Power Systems, pp. 1–1, 2020. [4] “ENTSO-E Network Code forRequirements for Grid Connection Applicable to all Generators.” (2021, March 13). [5] “ENTSO-E Network Code COMMISSION REGULATION (EU) 2017/1485.” (2021, March 13). [6] “TechnicalReport on the events of 9 August2019.” (2021, March 13). [7] SVK, “svenska kraftnat-system-development-plan-2018-2027.” (2021, March 13). [8] Y. G. S. Abouzeid and H. Zhang, “A frequency response strategy for variable speed wind turbine based on a dynamic inertial response and tip-speed ratio control.,” Electrical Engineering volume 101, pp. 35–44, 2019. [9] N. W. Miller and J. J. Sanchez-gasca, “Modeling of ge wind turbine-generators for grid studies,” GE Energy, vol. 98, pp. 44–46, 2008. [10] L. Ruttledge and D. Flynn, “System-wide inertial response from fixed speed and variable speed wind turbines,” in 2011 IEEE Power and Energy Society General Meeting, pp. 1–7, 2011. [11] F. Hafiz and A. Abdennour, “Optimal use of kinetic energy for the inertial support from variable speed wind turbines,” Renewable Energy, vol. 80, pp. 629–643, 2015. [12] J. Brisebois and N. Aubut, “Wind farm inertia emulation to fulfill hydro-quÃ©bec’s specific need,” in 2011 IEEE Power and Energy Society General Meeting, pp. 1–7, 2011. [13] M. Asmine, C.-Ã. Langlois, and N. Aubut, “Inertial response from wind power plants during a frequency disturbance on the hydro-quebec system â event analysis and validation,” IET Renewable Power Generation, vol. 12, no. 5, pp. 515–522, 2018. [14] L. Saarinen, P. Norrlund, W. Yang, and U. Lundin, “Linear synthetic inertia for improved frequency quality and reduced hydropower wear and tear,” International Journal of Electrical Power Energy Systems, vol. 98, pp. 488–495, 2018. [15] Ha Thi Nguyen, Guangya Yang, A. H. Nielsen, and P. H. Jensen, “Frequency stability improvement of low inertia systems using synchronous condensers,” in 2016 IEEE International Conference on Smart Grid Communications (SmartGridComm), pp. 650–655, 2016.. 41.

(53) [16] J. Liu, Y. Miura, and T. Ise, “Comparison of dynamic characteristics between virtual synchronous generator and droop control in inverter-based distributed generators,” IEEE Transactions on Power Electronics, vol. 31, no. 5, pp. 3600–3611, 2016. [17] J. Alipoor, Y. Miura, and T. Ise, “Power system stabilization using virtual synchronous generator with alternating moment of inertia,” IEEE Journal of Emerging and Selected Topics in Power Electronics, vol. 3, no. 2, pp. 451–458, 2015. [18] J. Alipoor, Y. Miura, and T. Ise, “Distributed generation grid integration using virtual synchronous generator with adoptive virtual inertia,” in 2013 IEEE Energy Conversion Congress and Exposition, pp. 4546–4552, 2013. [19] J. Zhu, C. D. Booth, G. P. Adam, and A. J. Roscoe, “Inertia emulation control of VSC-HVDC transmission system,” APAP 2011 - Proceedings: 2011 International Conference on Advanced Power System Automation and Protection, vol. 1, pp. 1–6, 2011. [20] I. Serban, R. Teodorescu, and C. Marinescu, “Analysis and optimization of the battery energy storage systems for frequency control in autonomous microgrids, by means of hardware-in-the-loop simulations BT - 2012 3rd IEEE International Symposium on Power Electronics for Distributed Generation System,” pp. 374–379, 2012. [21] A. Fernández-Guillamón, E. Gómez-Lázaro, E. Muljadi, and Á. Molina-García, “Power systems with high renewable energy sources: A review of inertia and frequency control strategies over time,” Renewable and Sustainable Energy Reviews, vol. 115, no. November, 2019. [22] S. Golestan, J. M. Guerrero, and J. C. Vasquez, “Three-Phase PLLs: A Review of Recent Advances,” IEEE Transactions on Power Electronics, vol. 32, no. 3, pp. 1897–1907, 2017. [23] X.-Q. Guo, W.-Y. Wu, and H.-R. Gu, “Phase locked loop and synchronization methods for grid- interfaced converters: a review,” PRZEGLA¸D ELEKTROTECHNICZNY (Electrical Review), vol. 87, no. 4, pp. 33–2097, 2011. [24] P. Cossutta, M. P. Aguirre, M. A. Engelhardt, A. Cao, and M. I. Valla, “High speed fixed point DSOGI PLL implementation on FPGA for synchronization of grid connected power converters,” IEEE International Symposium on Industrial Electronics, pp. 1372–1377, 2014. [25] Royan, T. Andromeda, M. Facta, Hermawan, and I. Setiawan, “Comparison of SOGI-FLL with SOGI-PLL for Single-Phase Grid-Connected Inverters,” E3S Web of Conferences, vol. 125, no. 201 9, pp. 3–7, 2019. [26] W. Ma, S. Ouyang, and W. Xu, “Improved frequency locked loop based synchronization method for three-phase grid-connected inverter under unbalanced and distorted grid conditions,” Energies, vol. 12, no. 6, 2019. [27] Z. Luo, M. Su, Y. Sun, Z. Liu, and M. Dong, “First-order generalized integrator based frequency locked loop and synchronization for three-phase grid-connected converters under adverse grid conditions,” Journal of Power Electronics, vol. 16, no. 5, pp. 1939–1949, 2016. [28] A. Savitzky and M. J. Golay, “Smoothing and Differentiation of Data by Simplified Least Squares Procedures,” Analytical Chemistry, vol. 36, no. 8, pp. 1627–1639, 1964.. 42.

(54) [29] R. Schafer, “What Is a Savitzky-Golay Filter? [Lecture Notes],” IEEE Signal Processing Magazine, vol. 28, no. 4, pp. 111–117, 2011. [30] J. Luo, K. Ying, P. He, and J. Bai, “Properties of Savitzky-Golay digital differentiators,” Digital Signal Processing: A Review Journal, vol. 15, no. 2, pp. 122–136, 2005. [31] R. I. A. Yazdani, Voltage-Sourced Converters in Power Systems : Modeling, Control, and Applications. Wiley-IEEE Press, 1 ed. ed., February 2010. [32] M. Fregelius and U. Lundin, “Hardware implementation of a synthetic inertia system for grid stability,” in 2019 8th International Conference on Renewable Energy Research and Applications (ICRERA), pp. 186–190, 2019. [33] J. Abrahamsson, J. J. Pérez-Loya, M. Fregelius, F. Evestedt, J. Bladh, and U. Lundin, “Magnetic thrust bearing for a 10 mw hydropower generator with a kaplan turbine,” Hydro 2018, 2018. [34] U. Lundin, F. Evestedt, J. Abrahamsson, J. Pérez, M. Fregelius, and J. K. Nøland, “Start of a synchronous motor using rotor field polarity inversion and rotor back-emf sensing,” in 2020 International Conference on Electrical Machines (ICEM), vol. 1, pp. 338–344, 2020. [35] A. Parwal, M. Fregelius , P. Almeida, O. Svensson, I. Temiz, J. Oliveira, C. Boström, and M. Leijon, “A comparative analysis of linear and nonlinear control of wave energy converter for a force control application,” International Marine Energy Journal, vol. 2, pp. 39–50, Jun. 2020. [36] A. Parwal, M. Fregelius, D. C. Silva, T. Potapenko, J. Hjalmarsson, J. Kelly, I. Temiz, J. G. de Oliviera, C. Boström, and M. Leijon, “Virtual synchronous generator based current synchronous detection scheme for a virtual inertia emulation in smartgrids,” Energy and Power Engineering, vol. 11, pp. 99–131, March. 2019. [37] A. Parwal, M. Fregelius, J. Leijon, M. Chatzigiannakou, O. Svensson, I. Temiz, C. Boström, J. G. de Oliveira, and M. Leijon, “Experimental test of grid connected vsc to improve the power quality in a wave power system,” in 2018 5th International Conference on Electric Power and Energy Conversion Systems (EPECS), pp. 1–7, 2018. [38] A. Parwal, M. Fregelius, I. Temiz, M. Göteman, J. G. de Oliveira, C. Boström, and M. Leijon, “Energy management for a grid-connected wave energy park through a hybrid energy storage system,” Applied Energy, vol. 231, pp. 399 – 411, 2018. [39] A. Parwal, M. Fregelius, J. Leijon, M. Chatzigiannakou, O. Svensson, E. Strömstedt, I. Temiz, J. G. de Oliveira, C. Boström, and M. Leijon, “Grid integration and a power quality assessment of a wave energy park,” IET Smart Grid, vol. 2, pp. 625–634(9), December 2019. [40] P. J. Vogler-Finck and W.-G. FrÃ¼h, “Evolution of primary frequency control requirements in great britain with increasing wind generation,” International Journal of Electrical Power Energy Systems, vol. 73, pp. 377–388, 2015.. 43.

(55)

(56)

No results found