Reference Power Point Tracking Reference Power Point Tracking of the Inertial Storage System Connected to the Electrical Grid

More and more the world is converging to the exploitation of electrical energy; As a result, electricity consumption continues to increase in considerable quantities. Nowadays, most of electrical networks are powered from power stations that use fossil-type primary energy. This energy on the one hand is limited in stocks, on the other pollutes the atmosphere, negative alteration of the climate namely the release of greenhouse gases. This will result in a consequent rise in temperature of the planet earth. This fact affects human health and ecological balance. The ideal would be to minimize the use of fossil energy, contribute to the purification of the environment and meet the high demand for electricity. This involves the integration of renewable resources, including unlimited human-scale stocks, into the electricity grid.

Various contributions and studies related to wind energy systems, with a focus on control techniques, fault management, optimization, and energy storage are presented in this work. The studies in references [1-4] mention various applications and methodologies related to wind energy systems. In the study of Hassani et al. [1] presented the application of a control technique to double the nominal power generated by the wind turbine. The capacity and fault management in an integrated micro-wind conversion system are improved via adaptive control [2]. The optimization criterion in order to minimize the cost of energy while taking into account the level of reliability required by the consumer is proposed by Djamila et al. [3]. Similarly, Mebarki et al. [4] proposed a systematic methodology to optimize the costs of an offshore wind farm based on a supervisory level frame work. These contributions present an attractive and alternative solution to fossil resources. However, their development is slowed down and considered as negative charges and does not contribute to system services or the protection of the electrical grid. This is first due to their primary sources which stochastic nature; therefore, do not participate in system services.

Otherwise, their rates of integration into the grid are limited to 30% of the capacity of the grid [5]. From these considerations it emerges that the stability of the network does not only depends on the reinforcement in electrical energy but on the quality of the power injected into the grid. This state should change and consciously thought to integrate the means of storage in the intermittent resources, which can make it acquire the smoothing property of the power injected into the network, and therefore, crossing the penetration barrier to the network and participate in the system service. Indeed, the integration of the storage system contributes to the valorization of intermittent and renewable energy production systems [6].

Many works are developed on energy storage systems; among these, there is the electromechanical storage, generally integrated in generators whose primary energy is stochastic. The aim of integrating this type of storage into generating systems is, in one hand, to increase the penetration rate of the generators to the micro-networks [7, 8] and, in the other hand, to participate in the system service and to improve the quality of the injected power in the grid [9-11]. Therefore, valorize intermittent resources and protect the network against overloads. Ayodele and Ogunjuyigbe [8] have shown the importance of integrating energy storage systems, particularly flywheels, into intermittent energy production units, with a view to reducing intermittency and making the most of renewable resources. But there is no mention of power losses in converters when power is exchanged between the grid and the storage system.

In order to improve the performance of these storage systems, several strategies and control techniques have been applied. Authors have tried to improve the performance of inertial storage systems, taking several steps. All the studies in references [12-15] mention challenges such as errors in current measurement, instability risks, and difficulties in identifying controller parameters. This highlights the need for robust control strategies and accurate parameter identification methods. For example, a control strategy of the synchronous machine driving a flywheel is presented, in order to improve the performance of the inertial storage system, nevertheless, the error during the current measurement increases after the integral, in addition, the speed estimation method using the waveforms is no longer credible [12]. A control technique is used to emulate a flywheel, which is made from the DC machine [13]. A flywheel is driven by the permanent magnet machine, which is controlled by means of a direct frequency matrix-converter [14]. However, for short operating periods there is an instability risk due to errors in the measurement of the stator currents. A new algorithm is proposed be Gamboa et al. [15] to control high-speed flywheels controlled by PID controllers, but, the application of this algorithm, presents difficulties to identify the parameters of the inverter with this controller, which makes sensitive the system control.

Talebi et al. [16] proposed the study of a variable speed wind induction generator associated to a flywheel energy storage system using direct torque control strategy through only simulations. Similarly, low speed PMSG based flywheel energy storage system is integrated with Direct-Drive (DD) variable speed PMSG based wind energy conversion system [17]. Or a flywheel driven by the squirrel-cage induction machine, for which a direct torque control (DTC) and vector modulation are applied to improve the performance of the system [18].

The possibility and interest of grouping inertia flywheels in parallel, and driving them by permanent magnet synchronous motors, has been treated by Xu et al. [19]. Aissou et al. [20] has been shown the interest of replacing mechanical bearings by magnetic ones, which lies in losses reduction. It is noticed there is a lack of mention of power losses in converters during power exchange between the grid and storage systems in some previous studies [21-23]. Considering converter losses is crucial for accurate assessments and system optimization.

In order to value inertial storage, an overview of flywheel technology and previous projects is presented in indeed [21], the work is dedicated to minimizing losses in flywheel design, without taking into account losses in the static converters involved in power exchange between the flywheel and the electrical grid. Bolund et al. [22] have demonstrated the possibility to compensate the reactive power, by integrating an inertial storage system in the wind turbine conversion system. It can be also found by Aissou et al. [23] that the role played by the inertial storage system associated with wind generators connected to the grid in. In the works above-cited, the reference of the power to be stored and restored is determined from the electromagnetic power. This method does not take into account the power lost in the static and dynamic converters. Therefore, the real power involved between the grid and the storage system is not reached, the correction of the latter is essential in order to stabilize the electrical grid [24].

To overcome this constraint, a new tracking algorithm for the storage power is proposed in this paper. The inertial storage system is made up with a flywheel and permanent magnet synchronous machine injecting power into the grid through power electronic interface. The proposed model will be implemented and simulated in Matlab/Simulink environment. The obtained results are commented and confirmed by comparison with experimental results. The real implementation is made in the laboratory with an emulated flywheel with a DC motor, for which the torque is programmed via a TMS 320F240-type DSP. Permanent magnet machine is connected to the grid through two back-to-back converters. The VSI are controlled by means of RTI1005-type DSPACE. Three experimental tests are made as follows:

-A test with a desired power profile to validate the operation of the system studied in motor /generator modes;

-A test in motor mode, in which the measured power is the one from the grid side;

-A test in motor mode, in which the measured power is the one from the machine side.

The results will be discussed and compared in order to show the effectiveness of the RPPT algorithm and the importance of operating with grid side power.

The bidirectional power converter consists of two conventional pulse width modulated (PWM) voltage source converters (VSC) [26, 27]. The first one is a rectifier while the second is an inverter. These two converters share the same DC-link (Figure 3).

3.png

Figure 3. The back-to-back rectifier-inverter converter

3.1 Rectifier modeling

The relation between AC and DC sides of the rectifier is given by the following relation.

$\left[\begin{array}{l}V_{d c}^{+} \\ V_{d c}^{-}\end{array}\right]=\left[\begin{array}{lll}S_1 & S_3 & S_5 \\ S_2 & S_4 & S_6\end{array}\right]\left[\begin{array}{l}v_a \\ v_b \\ v_c\end{array}\right]$      (13)

where, S_i is a connection function of associated with each switch in the rectifier (i=1,2,…,6). The DC-link voltage is obtained as:

$V_c=V_{d c}^{+}-V_{d c}^{-}$                       (14)

In the same way we can express DC current $i_{\text {rec}}$ according to the AC currents as:

$i_{r e c}=\left[\begin{array}{lll}S_1 & S_3 & S_5\end{array}\right]\left[\begin{array}{l}i_a \\ i_b \\ i_c\end{array}\right]$                    （15）

3.2 Filter modeling

The filter used in this paper is a low pass. It is represented in Figure 4.

4.png

Figure 4. Presentation of a low-pass filter

The model of the filter is defined by the following system of equation.

$\begin{gathered}i_{\text {rec }}-i_{\text {inv }}=C \frac{d v_c}{d t} \\ V_C(t)=V_C(0)+\frac{1}{C} \int_{t_1}^{t_2}\left(i_{\text {rec }}-i_{\text {inv }}\right) d t\end{gathered}$                   （16）        

The capacitor C represents the input DC-voltage for the inverter input, in order to supply reactive energy to the machine, and to absorb the negative current restored by the load.

3.3 Inverter modeling

The voltage inverter ensures the conversion of continuous energy to alternative current (DC/AC). Output voltages are governed by Eq. (17).

$\left[\begin{array}{l}v_A \\ v_B \\ v_C\end{array}\right]=\frac{V_c}{3}\left[\begin{array}{ccc}2 & -1 & -1 \\ -1 & 2 & -1 \\ -1 & -1 & 2\end{array}\right]\left[\begin{array}{l}S_7 \\ S_9 \\ S_{11}\end{array}\right]$                            （17）

where, S_i is a function of connection associated with each switch of the inverter (i=7, 8, …, 12).

The two converters, i.e., rectifier and inverter are PWM controlled. These converters consist of three arms formed with electronic switches chosen essentially according to the power and the commutation frequency. Each arm has two complementary power components provided with freewheel diode. This diode ensures the continuity of the current in the machine once the switch is open.

This section is dedicated to elaborate a tracking algorithm allowing to find the speed reference $\Omega_{r e f}$ which permits to track a given reference of grid-side power $P_D$ set by power system operator in order to adjust the grid frequency.

The relation between the mechanical power $P_{\text {mec }}$ and gridside one $P_g$ is given by the relation Eq. (20).

$P_{m e c}=P_g+\Delta_P$                       (20)

where, $P_{\text {mec }}$ is the algebraic value of the mechanical power. It is considered that $P_m>0$ once the machine is under motor mode and $P_{\text {mec }}<0$ if the machine is under generator mode. The same is applied for the grid-side power. $\Delta_P$ is power losses of the machine and the converters and it is also algebraic. $\Delta_{\mathrm{P}}<$ 0 if the machine is under motor mode, and $\Delta_{\mathrm{P}}>0$ if the machine is under generator mode.

It is given that:

$P_{\text {mec }}=T_r \Omega$                     (21)

The last relation implies:

$P_{m e c}=\Omega\left(T_{e m}-T_f-J \frac{d \Omega}{d t}\right)$                    (22)

Therefore,

$P_{m e c}=P_{e m}-\Omega\left(T_f+J \frac{d \Omega}{d t}\right)$                   (23)

Considering Eq. (20), the relation Eq. (23) becomes like in Eq. (24).

$P_g=P_{e m}-\Omega\left(T_f+J \frac{d \Omega}{d t}\right)-\Delta_P$                  (24)

It is given that:

$P_{e m}=\frac{3}{2} P \psi_r i_q \Omega$                 (25)

Its derivative time is:

$P_{e m}^{\cdot}=\frac{3}{2} P \psi_r \frac{d i_q}{d t} \Omega+\frac{3}{2} P \psi_r i_q \frac{d \Omega}{d t}$                 (26)

From the relation Eq. (24), it is given that:

$\dot{P}_g \cong P_{e m}^{\cdot}-\Delta_P-\dot{J}\left(\frac{d \Omega}{d t}\right)^2$                         (27)

By replacing Eq. (26) in Eq. (27), the relation Eq. (28) is obtained.

$\dot{P}_g \cong \frac{3}{2} P \psi_r \frac{d i_q}{d t} \Omega+\frac{3}{2} P \psi_r i_q \frac{d \Omega}{d t}-\Delta_P-\dot{J}\left(\frac{d \Omega}{d t}\right)^2$                          (28)

Introducing the tracking error $e_P$ defined as in (29).

$e_P=P_g-P_D$                          (29)

Its time derivative is given in Eq. (30).

$\dot{e}_P=\frac{3}{2} P \psi_r \frac{d i_q}{d t} \Omega+\frac{3}{2} P \psi_r i_q \frac{d \Omega}{d t}-\Delta_P-\dot{J}\left(\frac{d \Omega}{d t}\right)^2$                         (30)

The following Lyapunov is selected as given in the relation Eq. (31).

$V_P=\frac{1}{2} e_P{ }^2$                                   (31)

Its time derivative is given by (32).

$\dot{V_P}=\dot{e_P} e_P$                                   (32)

If the speed gradient is set as in Eq. (33); the quantity $\dot{V}_P$ becomes in Eq. (34). $\mu$ is a positive coefficient.

$\frac{d \Omega}{d t}=\frac{d \Omega_{r e f}}{d t}=-\mu {sign}\left(i_q\right) {sign}\left(e_P\right)$                                   (33)

$\begin{gathered}\dot{V}_P=\left(\frac{3}{2} P \psi_r \frac{d i_q}{d t} \Omega-\Delta_P-\dot{J}\left(\frac{d \Omega}{d t}\right)^2\right) e_P \\ -\frac{3}{2} \mu P \psi_r\left|i_q\right|\left|e_P\right|\end{gathered}$                                   (34)

The quantity $\dot{V}_P$ is negative if:

$\left|\frac{3}{2} P \psi_r \frac{d i_q}{d t} \Omega-\Delta_P-\dot{J}\left(\frac{d \Omega}{d t}\right)^2\right|>\frac{3}{2} \mu P \psi_r\left|i_q\right|$                      (35)

Under that condition it is obtained that: $e_P \rightarrow 0$

From the discretized formula of Eq. (33) the relation Eq. (36) is obtained.

$\begin{gathered}\Omega_{r e f}(k+1)=\Omega_{r e f}(k) -\mu T_d \operatorname{sign}\left(P_g(k)\right) \operatorname{sign}\left(P_g(k)-P_D(k)\right)\end{gathered}$                      (36)

$T_d$ is the discretization time.

As consequence, the speed reference allowing ensuring the tracking of grid side power to its reference is given by the relation Eq. (36). Its convergence is ensured if the coefficient $\mu$ satisfies the condition inequality (35).

The tracking algorithm proposed is described by the following steps:

1-From the profile of the desired power $P_D(k)$;

2-The initial rotation speed is set, and the power $P_g(k)$ is measured;

3-The signs of $P_D(k)$ and $\left(P_g(k)_{\text {mes }}-\left|P_D(k)\right|\right)$ are determined;

4-$\quad \Omega_{r e f}(k+1)=\Omega_{r e f}(k)-\mu T_d {sign}\left(P_g(k)\right) {sign}\left(P_g(k)-\right.$ $\left.P_D(k)\right)$

The proposed flowchart of the optimization algorithm is illustrated in Figure 5.

5.png

Figure 5. Flowchart of the RPPT algorithm

We recall the electrical and mechanical models representing the dynamics of the permanent magnet machine.

$\left\{\begin{array}{l}\frac{d i_d}{d t}=-\frac{R_s}{L_d} i_d+P \Omega \frac{L_q}{L_d} i_q+\frac{V_d}{L_d} \\ \frac{d i_q}{d t}=-\frac{R_s}{L_q} i_q-P \Omega \frac{L_d}{L_q} i_d-P \Omega \frac{\psi_r}{L_q}+\frac{V_q}{L_q} \\ \frac{d \Omega}{d t}=\frac{3 P\left(L_d-L_q\right) i_d+P \psi_r}{2 J} i_q-\frac{T_r}{J}-\frac{f}{J} \Omega\end{array}\right.$                       (37)

7.1 Regulator synthesis by sliding mode of speed

The relative degree is r=1, the surface is:

$S(\Omega)=\Omega^*-\Omega$                       (38)

The derivative of the surface S(Ω) is given by Eq. (39).

$\dot{S}(\Omega)=\dot{\Omega}^*-\dot{\Omega}$                          (39)

Substituting the derivative of the speed by its value, we obtain:

$\left\{\begin{array}{l}\dot{S}(\Omega)=\dot{\Omega}^*-\frac{3 P\left(L_d-L_q\right) i_d+P \psi_r}{2 J} i_q+\frac{T_r}{J}+\frac{f}{J} \Omega \\ i_q=i_{q e q}+i_{q n}\end{array}\right.$                          (40)

During the slip mode and its steady state, we have $S(\Omega)=$ $0, \dot{S}(\Omega)=0$ et $i_{q n}=0$.

From which we draw the expression of the equivalent component $i_{q e q}$.

$i_{q e q}=\frac{2 J \dot{\Omega}^*+2\left(\dot{T}_r+f \Omega\right)}{3 P\left(L_d-L_q\right) i_d+P \psi_r}$                          (41)

During the convergence mode, the derivative of the Lyapunov equation must be negative.

$S(\Omega) . \dot{S}(\Omega)<0$                          (42)

Substituting Eq. (41) into Eq. (40) we obtain Eq. (43):

$\dot{S}(\Omega)=-\frac{3}{2}\left[\frac{P\left(L_d-L_q\right) i_d}{J}+\frac{P \psi_r}{J}\right] i_{q n}$                          (43)

The non-linear control is:

$i_{q n}=K_{\Omega} \operatorname{sign}(S(\Omega))$                          (44)

where, $K_{\Omega}$ is positive gain.

7.2 Synthesis of control by sliding mode current quadrature

The surface chosen for the current is:

$S\left(l_q\right)=l_q^*-l_q$                          (45)

The derivative of the surface  $S\left(i_q\right)$ is:

$\dot{S}\left(l_q\right)=\dot{l}_q^*-i_q$                          (46)

$\left\{\begin{array}{l}\dot{S}\left(l_q\right)=l_q^*+\frac{R_s}{L_q} l_q+P \Omega \frac{L_d}{L_q} l_d+P \Omega \frac{\psi_r}{L_q}-\frac{V_q}{L_q} \\ V_q=V_{q e q}+V_{q n}\end{array}\right.$                          (47)

During the slip mode and its steady state, we have $S\left(l_q\right)=$ $0, \dot{S}\left(\mathrm{l}_{\mathrm{q}}\right)=0$ and $V_{q n}=0$.

From which we draw the expression of the equivalent component $V_{\text {qeq }}$.

$V_{q e q}=\left(l_q^*+\frac{R_s}{L_q} l_q+\frac{L_d}{L_q} P \Omega l_d+P \Omega \frac{\psi_r}{L_q}\right) L_q$                          (48)

Substituting Eq. (12) into Eq. (11) we obtain:

$\begin{gathered}\dot{S}\left(l_q\right)=\frac{-1}{L_q} V_{q n} \\ V_{q n}=K_q \operatorname{sign}\left(S\left(l_q\right)\right)\end{gathered}$                          (49)

where, $K_q$ is positive gain.

7.3 Syntheses of the regulator by sliding mode of the direct current

The surface is that of the current control $l_d$. It is described by:

$S\left(l_d\right)=l_d^*-l_d$                          (50)

The derivative of the surface $S\left(l_d\right)$ is:

$\dot{S}\left(l_d\right)=\dot{l}_d^*-\dot{l}_d$                          (51)

$\left\{\begin{array}{l}\dot{S}\left(l_d\right)=\dot{l}_d^*+\frac{R_s}{L_d} l_d-P \Omega \frac{L_q}{L_d} l_q-\frac{V_d}{L_d} \\ V_d=V_{d e q}+V_{d n}\end{array}\right.$                          (52)

During the slip mode and steady state, we have $S\left(l_d\right)=0$, $\dot{S}\left(l_d\right)=0$ and $V_{d n}=0$.

From which we draw the expression of the equivalent component $V_{ {deq }}$.

$V_{d e q}=\left(l_d^*+\frac{R_s}{L_d} l_d-\frac{L_q}{L_d} P \Omega l_q\right) L_d$                          (53)

Substituting Eq. (18) into Eq. (52), we obtain expression Eq. (54).

$\dot{S}\left(l_d\right)=\frac{-1}{L_d} V_{d n}$                          (54)

$V_{d n}=K_d {sign}\left(S\left(l_d\right)\right)$                          (55)

where, $K_d$ is positive gain.

7.4 Calculation parameters K_Ω, K_q  et K_d

These parameters are calculated for:

-Limit current to acceptable values for the maximum torque;

-Ensure the speed of convergence;

-Impose dynamics in convergence and slip mode.

7.5 Calculation of K_Ω

The convergence condition $S(\Omega) \cdot \dot{S}(\Omega)<0$ is ensured if:

(a) $Si\,S(\Omega)>0$ et $\dot{S}(\Omega)<0$

$\Omega^*-\frac{3 P\left(L_d-L_q\right) i_d+P \psi_r}{2 J} K_{\Omega}+\frac{T_r}{J}+\frac{f}{J} \Omega<0$                          (56)

$K_{\Omega}>\frac{2 J \dot{\Omega}^*+2 T_r+2 f \Omega}{3 P\left(L_d-L_q\right)+P \psi_r}$                          (57)

(b) ${Si}\,S(\Omega)<0$ et $\dot{S}(\Omega)>0$

$K_{\Omega}>-\frac{2 J \dot{\Omega}^*+2 T_r+2 f \Omega}{3 P\left(L_d-L_q\right)+P \psi_r}$                          (58)

Inequality (57) and inequality (58) take the expression locating the parameter $K_{\Omega}$.

$K_{\Omega}>\left|-\frac{2 J \dot{\Omega}^*+2 T_r+2 f \Omega}{3 P\left(L_d-L_q\right)+P \psi_r}\right|$                          (59)

7.6 Calculation of K_q

The convergence condition $S\left(l_q\right) \cdot \dot{S}\left(l_q\right)<0$ is ensured if:

(a) $Si  \,S\left(l_a\right)>0$ et $\dot{S}\left(l_q\right)<0$

$\dot{l}_q^*+\frac{R_s}{L_q} l_q+P \Omega \frac{L_d}{L_q} l_d+P \Omega \frac{\psi_r}{L_q}-\frac{K_q}{L_q}<0$                          (60)

$K_q>L_q\left(\dot{l}_q^*+\frac{R_s}{L_q} l_q+P \Omega \frac{L_d}{L_q} l_d+P \Omega \frac{\psi_r}{L_q}\right)$                          (61)

(b) ${Si}\, S\left(l_q\right)<0$ et $\dot{S}\left(l_q\right)>0$

$\dot{l}_q^*+\frac{R_s}{L_q} l_q+P \Omega \frac{L_d}{L_q} l_d+P \Omega \frac{\psi_r}{L_q}+\frac{K_q}{L_q}>0$                          (62)

$K_q>-L_q\left(i_q^*+\frac{R_s}{L_q} l_q+P \Omega \frac{L_d}{L_q} l_d+P \Omega \frac{\psi_r}{L_q}\right)$                          (63)

From inequality (61) and inequality (63) we deduce the expression locating the parameter K_q.

$K_q>\left|-L_q\left(i_q^*+\frac{R_s}{L_q} l_q+P \Omega \frac{L_d}{L_q} l_d+P \Omega \frac{\psi_r}{L_q}\right)\right|$                          (64)

7.7 Calculation of K_d

The convergence condition $S\left(l_d\right) \cdot \dot{S}\left(l_d\right)<0$ is ensured if:

(a) $Si \,S\left(l_d\right)>0$ et $\dot{S}\left(l_d\right)<0$

$\dot{l}_d^*+\frac{R_s}{L_d} l_d-P \Omega \frac{L_q}{L_d} l_q-\frac{K_d}{L_d}<0$                          (65)

$K_d>L_d\left(i_d^*+\frac{R_s}{L_d} l_d-P \Omega \frac{L_q}{L_d} l_q\right)$                          (66)

(b)  $Si\,S\left(l_d\right)<0$ et $\dot{S}\left(l_d\right)>0$

$i_d^*+\frac{R_s}{L_d} l_d-P \Omega \frac{L_q}{L_d} l_q+\frac{K_d}{L_d}>0$                          (67)

$K_d>-L_d\left(l_d^*+\frac{R_s}{L_d} l_d-P \Omega \frac{L_q}{L_d} l_q\right)$                          (68)

The parameter K_d is located from inequality (66) and inequality (68) in inequality (69).

$K_d>\left|-L_d\left(i_d^*+\frac{R_s}{L_d} l_d-P \Omega \frac{L_q}{L_d} l_q\right)\right|$                          (69)

The block diagram of vector control by sliding mode is presented in Figure 7.

7.png

Figure 7. Block diagram of vector control by sliding mode

9.1 Presentation of the test bench

The studied system is implemented in the electrical engineering laboratory (G2eLAB) in Grenoble, France (Figure 13).

13.png

Figure 13. The physical components of the real-time test bench

The generation of flywheel torque is realized by a direct current machine. The flywheel with inertia $J_v$ is emulated by a DC machine, by adjusting the electromagnetic torque reference. According to the mechanical equation of the shaft direct current machine.

$J \frac{d \Omega}{d t}=T_{e m}+T_r$                         (70)

By imposing the electromagnetic torque reference:

$T_{e m}=T_{e m_{-} r e f}=\left(J-J_v\right) \frac{d \Omega}{d t}$                         (71)

Substituting Eq. (71) in Eq. (70), we obtain the mechanical equation that governs the mechanical operation of the flywheel.

$J_v \frac{d \Omega}{d t}=T_r$                         (72)

Test bench consists of two essential parts:

1-(DC) whose torque is programmable via a DSP type TMS320F240. The whole is controlled by a TESTPOINT interface.

2-The machine: The machine used is of the permanent magnet synchronous type for which the stator is connected to the network via a power Bay which contains the AC/DC/AC power electronics interface, a connection transformer and the LC filters. The control of the whole is ensured by a DSPACE of type RTI1005.

We find in the complete bench the following elements:

-Mechanical part consisting of two machines with rigid coupling, the MCC which emulates the torque of the flywheel brought back to the shaft of the MS which is the reversible machine.

-The power bay associated with the DC motor consisting of a four quadrant chopper and its control, a DSP type TMS320F240, different sensor boards.

-The power bay associated with the synchronous motor: it contains two two-stage three-phase voltage inverters, a DSPACE system with a power PC, different sensor boards.

The network voltage before transformer (at the connection point of the synchronous machine to the EDF network) is about Veff=127V, 50Hz and the characteristics of the different components (inverter, grid, DC motor, PMSM) are given in the Tables 1-3.

Table 1. Bench characteristics inverter

Table 2. Internal parameters of the PMSM

Table 3. Parameters of the DC machine

9.2 Experimental results

The experimental results are presented below in three parts as follows:

9.2.1 Test with a desired power profile to validate the operation of the studied system as a motor/generator mode

In Figure 14, the machine rotation speed (Ω) follows exactly its reference (Ω_ref); It is of positive slope for the operation in storage mode and negative for the operating mode generator (restore). Also, the evolution of the quadrate component I_q is given, with positive or negative sign corresponding to the operating modes of the studied system, storage and destocking respectively. The active power waveform illustrates the direction of the power flow converted by the permanent magnet machine; the phase where the sign of the component of the current i_q is positive corresponds to the storage mode, conversely, it corresponds to the mode of power restitution. Figure 15 shows the evolution of the supply voltage of the machine and the current flowing through it; depending on the operating modes, phase difference between current voltage varies between the zero and π/2 during the storage mode and between π/2 and π during generator operating mode (destocking) (Figure 16).

14.png

Figure 14. Speed (Ω) and its reference (Ω_ref); current I_q(A) and active power (P_m) waveforms

15.png

Figure 15. Voltage and current waveforms during storage phase

16.png

Figure 16 Voltage and current phase discharge

Figure 17 shows the phase current and voltage waveforms where the machine is decoupled from the flywheel, therefore the power absorbed is lost in the machine.

17.png

Figure 17. Voltage and current where the machine is decoupled from the flywheel

9.2.2 Test in motor mode, the measured power on the grid side

In this test, the reference power RPPT is compared to that measured on the grid side. The results obtained are given in Figures 18 and 19. The rotational speed of the machine (Ω) which faithfully follows its reference (Ω_ref). The electrical network provides the desired active power (P_g), this is clearly shown by the perfect tracking of its reference (P_ref). It is also noted that the power absorbed by the stator is less than that provided by the network; this is due to losses in the static converter (Figure 18).

18.png

Figure 18. Speed and its reference Ω, Ω_ref, grid active power (P_g) and its reference RPPT_, P_ref; and machine active power P_m wave forms

Figure 19 shows voltage and stator current of the same phase (v_as), (i_as) respectively. These wave forms clearly show that the storage system operates in motor mode (storage phase).

19.png

Figure 19. Voltage and current waveforms during storage phase

9.2.3 Test in motor mode, the measured power on the machine side

In this test, the reference power RPPT is compared with that measured on the machine side. In the obtained results, it can notice that the speed of rotation of the machine (Ω) coincides with its reference (Ω_ref), the active electric power (P_m) absorbed by the motor which perfectly follows its reference (P_ref) and the active power provided by the network. It can be seen that the active power (P_g) supplied by the network is greater than that received by the machine. This is due to extra power provided by the network, and this power excess is dissipated in the power electronics bay. As a result, there is a risk of destabilizing the network (Figure 20).

20.png

Figure 20. Speed (Ω) and its reference (Ω_ref), grid active power (P_g), machine active power (P_m) and its reference RPPT (P_ref) waveforms

Figure 21 shows the voltage and the stator current of the same phase respectively, in which it can be seen clearly that the storage system works in motor mode (storage phase).

21.png

Figure 21. Voltage and current waveforms during storage phase