• No results found

3.5

Real-time simulation results

The first phase of design and validation of the controller was done using Matlab/Simulink, while HIL-RT approach was used to develop the final regula- tor and to validate its operation in a more realistic situation. An accurate model of the inverters and of the grid is simulated in the FPGA target of a National Instruments (NI) cRIO 9014 platform [121]. This model is a discrete time de- scription of the output LC filters of the inverters, the RL power lines, the load, and the grid-connection, as Fig. 3.5 shows. The discretization time is100 ns. The Pulse-Width Modulation (PWM), the inverter current and voltage regula- tors, and the proposed droop controllers are implemented in the FPGA targets of two NI GPICs [121]. NI cRIO and NI GPICs can interact because NI cRIO has some analog output ports that generate the signals that correspond to the inductor currents, the output voltages, and the output currents of the inverters. These analog quantities are sampled by the input analog ports of the NI GPIC which calculates the gate commands to modulate the inverters. Gate commands are given by the NI GPIC thanks to its output digital ports and they are read by the input digital ports of the NI cRIO platform. The basic scheme of the RT simulation setup is shown in Fig. 3.5. More details on this modeling and validation approach are in [62].

Figs. 3.6-3.9 report the RT simulation results for grid-connected operation and Fig. 3.10 the corresponding operating points of these simulations on the static droop characteristics. The operating point is indicated by the same letter in the Figs. 3.6-3.9 and Fig. 3.10. These figures show that the regulator can track its active and reactive power references in the range of frequency that is selected in the design phase, i.e. (3.22a). Figs. 3.6 and 3.7 show two DERs tracking null active and reactive power references at the minimum and max- imum frequencies of the grid voltage respectively, i.e. 49.5 Hz and 50.5 Hz. These two cases validate the choice of the saturation levels of the regulator Gp(s). It is possible to observe that the output current of inverter 1 is almost

zero in both cases.

The simulations of step variations of the active and reactive power refer- ences of the inverter 1 are reported in Figs. 3.8 (active power reference variation

G

Analog Outputs Analog Outputs Digital Inputs

Gate commands

NI GPIC Inverter 1 controller: current, voltage and droop controllers

NI GPIC Inverter 2 controller: current, voltage and droop controllers Gate commands Simulated

NI cRIO 9014 Power system network NI cRIO 9014 DER 1 + Electric lines + Load Main grid connection PCC Digital Inputs DER 2

Figure 3.5: RT-HIL platform organization [100]

Grid-connected at fgrid,min= 49.5 Hz vo1 io1 pm1 pm2 Null power references Pload= 2.4 kW a

Figure 3.6: RT simulation results of steady-state grid-connected operating mode atfgrid,min:

output voltage of inverter 1,vo1, in CH1 (256 V/div); output current of inverter 1, io1, in CH2

(12 A/div); measured active power of inverter 1, pm1, in CH3 (1024 W/div) and measured

3.5 Real-time simulation results 81 Grid-connected at fgrid,max= 50.5 Hz vo1 io1 pm1 pm2 Null power references Pload= 2.4 kW b

Figure 3.7: RT simulation results of steady-state grid-connected operating mode atfgrid,max:

output voltage of inverter 1,vo1, in CH1 (256 V/div); output current of inverter 1, io1, in CH2

(12 A/div); measured active power of inverter 1, pm1, in CH3 (1024 W/div) and measured

active power of inverter 2,pm2, in CH4 (1024 W/div); time with 20 ms/div [100]

from0 W to 2 kW) and 3.9 (reactive power reference variation from 0 VAR to 1 kVAR). The rise time is about 210 ms for both transients of Figs. 3.8 and 3.9. In all these cases, the resistive loadRLhas a nominal power of2.4 kW.

Grid-connected at fgrid= 50 Hz io1 pm1 pm2 Pload= 2.4 kW Null power references vo1 d c pref 1step change 2 kW 0 W

Figure 3.8: RT simulation results ofpref 1 step variation in grid-connected operating mode:

output voltage of inverter 1,vo1, in CH1 (256 V/div); output current of inverter 1, io1, in CH2

(12 A/div); measured active power of inverter 1, pm1, in CH3 (1024 W/div) and measured

active power of inverter 2,pm2, in CH4 (1024 W/div); time with 100 ms/div [100]

Grid-connected at fgrid= 50 Hz io1 qm1 qm2 Pload= 2.4 kW Null power references vo1 qref 1step change 1 kVAR

Figure 3.9: RT simulation results ofqref 1 step variation in grid-connected operating mode:

output voltage of inverter 1,vo1, in CH1 (256 V/div); output current of inverter 1, io1, in CH2

(12 A/div); measured reactive power of inverter 1, qm1, in CH3 (512 VAR/div) and measured

3.5 Real-time simulation results 83 −1.5 −1 −0.5 0 0.5 1 1.5 49.4 49.6 49.8 50 50.2 50.4 50.6

Characteristics for grid-connected mode simulations

Normalized power [ p.u.]

F requency [Hz] c d a b ps,max ps,min

Figs. 3.11-3.14 show the results for islanded operation and for the discon- nection of the main grid and Fig. 3.15 the operating points of the corresponding conditions on the static droop characteristics. The same operating point is indi- cated by the same letter in Figs. 3.11-3.14 and in Fig. 3.15. The transition from grid-connected to islanded operation is shown in Fig. 3.11. Zero active and re- active powers are generated by the two inverters before the disconnection and, after the transient, the load power is shared between the two energy sources, be- cause (3.19) is satisfied. It is possible to observe that the controller can achieve a smooth transition and that the frequency of the microgrid varies during this transient because the integral controllers saturate atps,min. A step variation of

the load resistance in islanded mode is reported in Fig. 3.12, showing the tran- sient time of about70 ms. In Fig. 3.13 there is a similar simulation that shows the power sharing failure. In this case, the relation (3.20) is satisfied after the transient and the two inverters do not share the load properly, but one inverter is still tracking its power reference. In Fig. 3.14, the failure of the power sharing is shown in a different way: the two inverters are working in grid-connected mode with opposite power references (±3 kW) and then a main grid discon- nection occurs. In this case, one inverter is absorbing an active power up to its rated power in order to show the worst case where the power sharing is not achieved. Indeed, in islanded operation the two inverters continue to provide active powers that are close to their references: inverter 1 is still tracking its reference, while inverter 2 is balancing the power that is required by the load.

3.5 Real-time simulation results 85

Main grid disconnection

io1

pm1

pm2

Null power references Pload= 2.4 kW f1 0 W 1.2 kW 50 Hz 49.39 Hz c e

Figure 3.11: RT simulation results of grid transition: frequency of the output voltage of inverter 1vo1,f1, in CH1 (0.6 Hz/div); output current of inverter 1, io1, in CH2 (3.2 A/div); measured

active power of inverter 1,pm1, in CH3 (1024 W/div); measured active power of inverter 2,

pm2, in CH4 (1024 W/div); time with 500 ms/div [100]

Islanded operation - Pload variation from 2.4 to 5.3 kW

pm1

pm2

Null power references

1.2 kW Frequency from 49.39 Hz to 49.36 Hz f e 2.7 kW io1 vo1

Figure 3.12: RT simulation results of load step variation in islanded operating mode: out- put voltage of inverter 1,vo1, in CH1 (192 V/div); output current of inverter 1, io1, in CH2

(12 A/div); measured active power of inverter 1, pm1, in CH3 (1024 W/div); measured active

Islanded operation - Ploadvariation from 5.3 to 2.4 kW pm1 2.7 kW Frequency from 49.36 Hz to 49.42 Hz f g io1 vo1 h 2 kW

Pref 1= 2 kW Pref 2=−1 kW Null reactive power references

pm2 400 W

Figure 3.13: RT simulation results of load step variation in islanded operating mode: out- put voltage of inverter 1,vo1, in CH1 (192 V/div); output current of inverter 1, io1, in CH2

(12 A/div); measured active power of inverter 1, pm1, in CH3 (1024 W/div); measured active

power of inverter 2,pm2, in CH4 (1024 W/div); time with 200 ms/div [100]

Main grid disconnection pm1

pm2 3 kW

Pref 1= 3 kW Pref 2=−3 kW

Null reactive power references Pload= 530 W i io1 j 50 Hz f 1 49.49 Hz 3 kW 3.15 kW (overcurrent) −3 kW −2.5 kW

Figure 3.14: RT simulation results of grid transition: frequency of the output voltage of inverter 1vo1,f1, in CH1 (0.6 Hz/div); output current of inverter 1, io1, in CH2 (20 A/div); measured

active power of inverter 1,pm1, in CH3 (1024 W/div); measured active power of inverter 2,

3.5 Real-time simulation results 87 −1.5 −1 −0.5 0 0.5 1 1.5 49.34 49.36 49.38 49.4 49.42 49.44 49.46 49.48 49.5 49.52

Normalized power [ p.u.]

F requency [Hz] h j g e ps,min i f Characteristics for islanded mode simulations

Related documents