This work investigates the possibility of using the energy stored in the active elements of a static power converter that transfers power from solar panels to the grid for auxiliary purposes. In particular, its propose the option of using the energy of the DC-Link capacitor to obtain more power than the solar panel can generate by itself. At cost of leaving the optimum point of maximum generation of the plant and this way to contribute with more energy to the network when it requires it for short intervals of time.
For this it is obtained the formulas that model the behavior of a solar panel, as a function of solar radiation and ambient temperature, a voltage source inverter and a LCL filter at the output. To determine the parameters is used the MathCad calculation program, is established all the equations according to input parameters in this way it can be obtained the dimensions of the components for different situations.
In addition, a control algorithm is presented capable of operating in two situations. The first is when the system is in normal mode, i.e. when the power consumed in the network is equal to or less than the maximum power generated by the photovoltaic panel. The second situation occurs when the power required by the grid is greater than the PV panel can generate.
The models are verified by Psim simulations in open and closed loop, with emphasis on the dynamics of the response and its stability. The results are verified by experimental tests carried out in the laboratory, using the components available in this laboratory.
The results show that it is possible to provide this energy, while the time during which it can be supplied depends on the size of the capacitor and the amount of energy needed.
One of the most important aspect when it comes to generate electrical energy is the utility frequency control. Frequency is modified every time there is a disparity between the power generated and the power consumed in the system.Β
Nowadays, in order to keep the utility frequency constant, the largest hydroelectric plants do not operate at maximum capacity, so if power demand increases they have an energy reserve to respond. The problem with this mechanism is that the operating dynamics of an hydroelectric power station is slow. Another alternative is to disconnect loads automatically until the frequency recovers.
The topology consists in a three phase inverter connected to a LCL filter, and to the grid.
For the control of the inverter two references are used that operate independently one from the other as shown in the Fig. The first one operates in normal mode, i.e. when the system does not demand more power than the photovoltaic panel can supply by itself and it is given by the MPPT. This control aims to obtain the maximum panel power using method P&O.
The second reference will come into operation when the system requires more power than the panel can supply and it will be given by the demanded power. Hence, the capacitor is discharged delivering additional energy for a short period of time.
It is important to emphasize that for the control the currents are measured at the inverter output and the grid voltage at each of its phases. Then, these measures are transformed to the rotating dq0 frame and these transformed variables are those that are used in the control loops.
Subsequently, the calculated modulators are converted to the stationary abc frame again.
The following hows the complete control diagram and how the 2 current references ππβ are generated, when the generation system operates normally and when more power is required than the system can provide using only the photovoltaic panels. In contrast, the reference for the reactive power ππβ is kept at zero because the main objective is to inject more active power to the system.
A feedforward strategy is used by the internal current loops to obtain a better response than the PI controllers, besides that it acts as a decoupling term.
During the simulation, the MPPT reference is used and in t = 2 the reference 2 is activated and reference 1 deactivated, i.e. more energy was injected to the grid. Then in the t = 2.02 the system returned to the MPPT reference. The results are shown below.
The figure shows how the capacitor gets discharged when more power its required by the grid, decreasing its voltage almost to 25 V. Naturally, the maximum power point tracking for the panel is lost.
The figure shows the power delivered to the grid. In t = 2 the power injected into the AC mains increases for nearly 20 ms.
The figure below shows the currents of the three phases at the AC network of the inverter and at time 2 (s) the reference is changed and more power is injected to the system accordingly the amplitudes increase. Then, in time 2.02 (s) the MPPT reference is returned and the currents undergo a small transient, until the capacitor is recharged.
The figure compares the active power delivered by the inverter for different dt (Time to supply extra power). As expected, while this period is shorter, the lower the power drop when returning to the MPPT, since the capacitor is less discharged.
The circuit that was implemented is shown in the figure below, the location of the current and voltage sensors within the physical circuit are indicated.Β
To avoid depending on solar conditions, we decided to use a solar panel emulator, available in the laboratory.
The control algorithm was programmed in C for a DSP TMS320F28335; using Code Composer Studio 6.1.1.
Then, the three-phase inverter was assembled:
Three Phase Inverter
Single leg
A card that converts the electrical signals to light was used to isolate the DSP from the inverter. This prevents noise from being generated in the control signal cables and protects the DSP against possible short circuits.
To control the system, 3 current and 3 voltage sensors were adapted, one for each phase. In addition, an extra pair of sensors for the solar panel was used.
Voltage Sensors
Current Sensors
The sensors output is a current signal while the DSP can only read voltage signals between 0 and 3V; therefore, in order to read the signals, the following card was used to adapt the signal.
For the grid connection, a three-phase Variac was used, which allowed to regulate the voltage to desired levels. In addition, for greater safety at the Variac output, transformers were connected to galvanically isolate the network.
Whole setup:
The most important data obtained from the set-up are: The currents of each phase and the DC link voltage.
First, the operation of the MPPT is checked, this is done using the emulator software. The point of maximum power is marked with an X while the point at which it is operating is marked with a circle; as seen in Figure 5.12 the panel would work close to the optimum (MPPT) assigned to the 39 V simulator; the overvoltage is due to sensor miscalibration. Where the error is 10%.
Figure 5.12
Figure 5.13
The waveforms measured by the sensors can then be observed on the oscilloscope in Figure 5.14. The current and voltage are in phase, as proposed in the control and circuit design. While in Figure 5.15 the inverter output currents of the 3 phases are shown.
Figure 5.14
Figure 5.15
It can be seen from the previous figures that the MPPT control and algorithm was successfully implemented. Due to the implemented algorithm, Perturb & observe the signals maintains a small oscillation. Furthermore, the exact control points are not reached due to the difficult calibration of the sensors.
Finally, it is verified that it is possible to get extra energy from the DC link capacitor for a short period of time at the cost of leaving the optimal generation point of the system.
From the work carried out, it can be concluded that it is possible to use the dc-link capacitor to inject power with minimum dynamics when the system requires it, this was demonstrated using a topology as close to reality as possible.
In addition, it was tested for different times during which the capacitor must supply power. From this it is concluded that the plant after having provided the extra energy, needs a time to recharge the capacitor and return to normal conditions.
It is possible to design a control capable of operating in 2 different modes and switching from one mode to another without stopping the plant.
The system supports the use of an LCL filter, which allows reducing the size of the components.
It can be expanded for higher capacity solar plants simply by increasing the capacitance of the capacitor, which does not imply losses in response speeds.
From the above it is concluded that solar panels are a more attractive and clean alternative, because the dynamics they present are smaller.