Design, modeling and simulation of a PV power plant

Universitat Polit`ecnica de Catalunya Escola T` ecnica Superior d’Enginyeria Industrial de Barcelona. ETSEIB Design, modeling and simulation of a PV power plant. Bachelor’s final project Autor: Pablo Poza Ferragut Supervisor: Oriol Gomis Bellmunt Junio 2020 Summary The rise of the photovoltaic energy during the last decades and mainly during the last years has many indicators; the fast growth of the global production capacity, the innovations related to the solar energy technologies or the continuous adaptation of the law code to fit a wider variety of scenarios and conditions. Additionally the recent improvements have given Large Scale PV power plant (LS-PVPP) the capability to work as baseload power plants relieving dirtier sources of energies. However one of the biggest disadvantages in front of conventional energy sources is the cost. In the last years PV technologies have achieved higher cost-effective ratio and now are able to compete against non-renewable production methods. For this reason the present work is centered in the development of a tool for integration of LS-PVPP in the actual energy distribution system that allows the user to optimize and study the design and functionality of the plant. A MATLAB based program and different functions have been developed to perform a power flow analysis. Different grid codes requirements have been analyzed for LS-PVPP. Approaches from different years are taken into account and the tendency of unifying some parts of these grid codes to achieve higher levels of energy share is discussed too. Examples of control implementation to meet these requirements are also included. The example used to test the Power flow equations (PFE) solver is composed of central inverters. According to [2] this configuration is the most cost-effective nowadays despite the string inverters are close and present a series of benefits related with the versatility and control. Contents 1 Prefase 8 1.1 Objective .................................... 8 1.2 Motivation ................................... 8 2 Power flow analysis 9 2.1 Procedure.................................... 9 2.2 Power flow equation basis . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.3 Formulation................................... 11 2.3.1 Bus admittance matrix . . . . . . . . . . . . . . . . . . . . . . . . 11 2.4 Limitations and considerations . . . . . . . . . . . . . . . . . . . . . . . . 12 2.5 Solutionmethods. ............................... 12 2.6 Definitions.................................... 13 2.6.1 Slack bus................................. 13 2.6.2 Load bus; PQ type. . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.6.3 Voltage controlled bus; PV type. . . . . . . . . . . . . . . . . . . . 14 2.7 The Newton-Raphson power flow solution. . . . . . . . . . . . . . . . . . . 14 2.7.1 Problem general form . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.7.2 Multi-Variable NRA . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.8 Load flows of the system . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 3 Validating the results 18 3.1 Considerations.................................. 18 3.1.1 Limitations ............................... 18 3.1.2 Distribution system limitations . . . . . . . . . . . . . . . . . . . . 19 1 3.1.3 Parameters under study. . . . . . . . . . . . . . . . . . . . . . . . . 19 3.2 Examplecase................................... 20 3.2.1 Perunitvalues ............................. 21 3.2.2 Initial values and suppositions . . . . . . . . . . . . . . . . . . . . 21 3.3 Result comparison versus MATPOWER . . . . . . . . . . . . . . . . . . . 23 3.3.1 First comparison: Active and reactive power demand variation with PQbuses................................. 23 3.3.2 Second comparison: Active and reactive power generation variation withPQbuses.............................. 26 3.3.3 Third comparision: Active power generation and grid voltage variationwithPVbuses........................... 28 3.4 Analysis of maximum and minimum voltage deviations and loses. . . . . . 30 3.4.1 Particular case: different values for each converter output . . . . . 30 4 Grid code requirements 33 4.1 Introduction................................... 33 4.2 Staticregulation ................................ 35 4.2.1 Power factor regulation . . . . . . . . . . . . . . . . . . . . . . . . 35 4.2.2 Power Curtailment . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 4.2.3 Active power reserves . . . . . . . . . . . . . . . . . . . . . . . . . 36 4.2.4 Voltage range and Control . . . . . . . . . . . . . . . . . . . . . . . 37 4.2.5 Remote Voltage Control . . . . . . . . . . . . . . . . . . . . . . . . 37 4.2.6 Frequency................................ 37 4.2.7 Flicker.................................. 38 4.3 Grid support; reactive power compensation . . . . . . . . . . . . . . . . . 38 5 Environmental impact analysis 42 5.1 Large scale PV power plant . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.1.1 Construction .............................. 43 5.1.2 Operation................................ 43 5.1.3 Decommission.............................. 44 5.1.4 Potential environmental consequences . . . . . . . . . . . . . . . . 44 5.1.5 Potential ecological consequences . . . . . . . . . . . . . . . . . . . 44 5.1.6 Transmission system impact . . . . . . . . . . . . . . . . . . . . . . 45 5.1.7 Conclusions............................... 45 2 6 Projects budget 47 6.1 Large Scale PV power plant . . . . . . . . . . . . . . . . . . . . . . . . . . 47 6.2 Ancillaryservices................................ 48 6.3 Introduction................................... 51 6.4 Scripts...................................... 51 6.5 First comparison power flow results . . . . . . . . . . . . . . . . . . . . . . 52 6.6 Second comparison power flow results . . . . . . . . . . . . . . . . . . . . 54 6.7 Third comparison power flow results . . . . . . . . . . . . . . . . . . . . . 55 3 List of Figures 2.1 Example of a 5 bus system with the impedance of the lines. . . . . . . . . 10 3.1 Relation between the Pdc and Vdc and main points to analyze. . . . . . . . 19 3.2 Example case with designated buses. . . . . . . . . . . . . . . . . . . . . . 20 3.3 Eleventh bus values (1st case) . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.4 Second bus values (1st case) . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.5 Seventh bus values (1st case) . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.6 Eleventh bus values (2nd case) . . . . . . . . . . . . . . . . . . . . . . . . 27 3.7 Second bus values (2nd case) . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.8 Seventh bus values (2nd case) . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.9 Eleventh bus values (3rd case) . . . . . . . . . . . . . . . . . . . . . . . . 29 3.10 Second bus values (3rd case) . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.11 Seventh bus values (3rd case) . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.12 Example scheme with voltage magnitude and phase values . . . . . . . . . 31 3.13 Example scheme with voltage magnitude and phase values . . . . . . . . . 31 3.14 Real and Imaginary losses changing the grid voltage from 0.9 to 1.1 . . . 32 4.1 Power installed in Spain from 2007 to 2018. . . . . . . . . . . . . . . . . . 34 4.2 Summary of the total energy managed by the ancillary services in the spanishpeninsula. ............................... 34 4.3 Reactive power requirements for Puerto Rico, South Africa and China, GermanyandRomania. ............................ 35 4.4 Power curtailment requirements . . . . . . . . . . . . . . . . . . . . . . . . 36 4.5 Summary of grid codes requirements for LS-PVPP . . . . . . . . . . . . . 38 4.6 Scheme of the connection points contextualized in the current example . . 39 4 4.7 From left to right; output PoC and grid PoC phasor diagrams when PF is settounity.................................... 40 6.1 Breakdown of ancillary services cost in the average final price of energy in thepeninsularsystem ............................. 49 6.2 Active and reactive power demand (1st case) . . . . . . . . . . . . . . . . 52 6.3 Active and reactive power generation (1st case) . . . . . . . . . . . . . . . 52 6.4 Imaginary and real losses (1st case) . . . . . . . . . . . . . . . . . . . . . . 53 6.5 Active and reactive power demand (2nd case) . . . . . . . . . . . . . . . . 54 6.6 Active and reactive power generation (2nd case) . . . . . . . . . . . . . . 54 6.7 Active and reactive power generation (2nd case) . . . . . . . . . . . . . . 54 6.8 Active and reactive power demand (3rd case) . . . . . . . . . . . . . . . . 55 6.9 Active and reactive power generation (3rd case) . . . . . . . . . . . . . . . 55 6.10 Active and reactive power generation (3rd case) . . . . . . . . . . . . . . . 55 5 List of Tables 3.1 Present work system parameters. . . . . . . . . . . . . . . . . . . . . . . . 21 3.2 Example’s parameters in pu. . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.3 Required parameters for each bus. . . . . . . . . . . . . . . . . . . . . . . 22 3.4 Required parameters for each bus. . . . . . . . . . . . . . . . . . . . . . . 22 3.5 Selected variables and ranges . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.6 Selected variables and ranges . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.7 Selected variables and ranges . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.8 Initial values for the converter output buses. . . . . . . . . . . . . . . . . . 30 5.1 Comparison of life cycle emissions for solar technologies and conventional carbon-intensive systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 6.1 PV plant CAPEX costs and current project budget . . . . . . . . . . . . . 48 6.2 PVplantOPEX ................................ 48 6 Acronyms DES Distributed energy sistems. 19, 36, 37 ENSTO-E European Network for Transmission System Operators for Electricity. 33 eSCR Equivalent short circuit ratio. 40 LS-PVPP Large Scale PV power plant. 1, 4, 34, 37–40, 42–45, 47 MPP maximum power point. 19 MPPT Maximum power point tracker. 30 NRA Newthon-Raphson algorithm. 14, 16, 17, 24, 26, 51 PF Power factor. 5, 35, 38, 40 PFC Power factor control. 39 PFE Power flow equations. 1, 23, 28, 34 PoC point of connection. 5, 38–41 RfG Network Code on Requirements for Grid Connection applicable to all Generators. 33 SCR Short circuit ratio. 39, 40 TSO Transmission system operator. 33, 36 7 Design, modeling and simulation of a PV power plant 14 2.6.2 Load bus; PQ type. This classification refers to the buses where the real and reactive power are specified. It is recommended to designate any bus with injected complex power as a load bus. The voltage magnitude and angle adapt to the situations created by the power injections. It has to be mentioned that the voltages have limitations to ensure smooth and secure voltage ranges for the devices connected to the grid. Usually these ranges go from 90 to 110% of the nominal voltage. 2.6.3 Voltage controlled bus; PV type. As its name describes the voltage is controlled adding a voltage controller module that sets the voltage in a narrow range that can be understood as constant. In addition the real power is also specified. Consequently the reactive power is a variable with a limitations with upper and lower bounds. In a real situation a PV bus always have a generator that meets the requirements to be a variable source of reactive power. 2.7 The Newton-Raphson power flow solution. 2.7.1 Problem general form In this section the Newton-Raphson algorithm is described; in the first part the presentation and general form of the problem are introduced. In the second part the matricidal power flow solution is discussed. In practice, the bus apparent net power Siis a parameter more accessible than the bus currents Iitherefore NRA reorganizes the equations following the next procedure. Regrouping terms the generic admittance expression is obtained: Yij =|Yij|∠θij =|Yij|cos θij +j|Yij |sin θij =Gij +jBij (2.12) Using the Gauss power flow (2.8) S∗ ican be expressed as: S∗ i=V∗ iIi=V∗ i n X j=1 (YijVj) (2.13) Now reorganizing equation (2.8) in correlation with the generic admittance reorganization (2.12) and the apparent power conjugate S∗ iequation (2.13) results the next expression: S∗ i=Pi−jQi= n X j=1 |YijViVj|∠(θij +δj−δi) (2.14) Finally 2nequations are obtained following the generic form of: Pi=|Vi|2Gii + n X j=1, j6=i |YijViVj|cos (θij +δj−δi) (2.15) Design, modeling and simulation of a PV power plant 15 Qi=|Vi|2Bii + n X j=1, j6=i |YijViVj|sin (θij +δj−δi) (2.16) The key idea behind the Newton-Raphson algorithm is to use a sequential linearization where function depending on xis equaled to zero. Starting from the initial guess every step an increment of ∆xis defined: f(ˆx) = 0 (2.17) ∆x(v)III =x−x(v)(2.18) Representing f(x) by a Taylor series where the components or 2nd order or higher are neglected. This can be done and performs well because the equation equals to zero. The expressions remains now as: f(ˆx) = f(x(v)) + df(x(v)) dx ∆x(v)(2.19) Finally it can be solved for ∆x(v)and start a new iteration setting the value of x(v+1) as the new one. ∆x(v)=−df(x(v)) dx −1 f(x(v)) (2.20) This algorithm has a quadratic convergence meaning the error decreases quickly when approaching the solution and the result depends on the initial guesses that can be set for the voltages. Two are considered; x(0) = 0 or x(0) =−1. The iteration stop when the absolute value of the function with the current x(n)is lower than a tolerance, set by the user. 2.7.2 Multi-Variable NRA The first change to notice is that in the multi-variable case the variables and functions ˆx, f(ˆx) and therefore the increment ∆xbecome vectors containing the parameters for each bus. In addition the derivative component is now a matrix known as the Jacobian. This matrix contains the partial derivative of each function f(ˆx) for every variable xn. J(x) =       ∂f1(x) ∂x1 ∂f1(x) ∂x2. . . ∂f1(x) ∂xn ∂f2(x) ∂x1 ∂f2(x) ∂x2. . . ∂f2(x) ∂xn . . .. . ..... . . ∂fn(x) ∂x1 ∂fn(x) ∂x2. . . ∂fn(x) ∂xn       (2.21) The system has a matrix structure therefore the operations must be performed in accordance. Following the solution scheme of previous sections the jacobian is inverted to solve for the increment ∆x. IIIv: represents the current step Design, modeling and simulation of a PV power plant 16 The NRA application to power flow need a reformulation for the power equations associated to each bus. Taking in to account the equations (2.12) and the conjugate of (2.14) and defining the next expressions: Vi=|Vi|ejθi=|Vi|∠θi ejθ = cos θ+jsin θ θij =θi−θj The following equation can be obtained: Si= n X j=1 |Vi||Vj|ejθij (Gij −jBij) (2.22) And resolving into the real and imaginary parts the equations used in the algorithm can be found. Pi= n X j=1 |Vi||Vj|(Gij cos θij +Bij sin θij) = PGi −PDi (2.23) Qi= n X j=1 |Vi||Vj|(Gij sin θij −Bij cos θij) = QGi −QDi (2.24) With the first bus designated as the slack bus and setting its voltage magnitude and angle the rest of the voltages and phases are found following the next scheme: x=           θ2 . . . θn |V2| . . . |Vn           f(x) =           P2(x)−PG2+PD2 . . . Pn(x)−PGn +PDn Q2(x)−QG2+QD2 . . . Qn(x)−QGn +QDn           At the beginning of the iteration the number of steps is set to 0, v= 0 and the initial guesses are introduced. Each step the number is increasing in one, v=v+1 The stopping criteria is the same as seen before for a given tolerance, . Design, modeling and simulation of a PV power plant 17 PV cases As long as the PV buses are fixed in voltage magnitude there’s no need to include the unknown in the system or write the reactive power balance equations; it varies to maintain the fixed voltage within a reasonable limits. However those can be included writing the voltage constraint as |Vi|−Visetpoint = 0. If they are not included in the system the script used in this work contains a simple limit violation test. It checks if the limits introduced in the bus data are violated. If the bounds are overpassed the program displays a message indicating which bus is having issues and the magnitude of these. 2.8 Load flows of the system Once the NRA converged the Voltage magnitudes and angles are known and the line flows and losses are obtained. First of all the Line Current Flows are calculated in each branch using Ohm’s law in both directions of the branch. The resulting expressions are shown next: The from bus-to bus current expression is obtained: Iij =−(Vi−Vjabranch)Yij /a2 branch +bbranch/a2 branchVi(2.25) and the reverse direction expression results in: Iji =−(Vj−Viabranch)Yji/a2 branch +bbranchVj(2.26) Where ais the tap setting and is set to 1 for all simulations. This tap setting is a functionality found in some transformers. It allows the transformer to have variable turn ratios to keep the changes of the primary side. The ground admittance is represented by b. In this work the ground admittance is the half of the susceptance present in any of the two buses conforming the line. Secondly, the Line Power Flows in MVA and the line losses are obtained using the apparent power equations: Sij =ViI∗ ij ∗Sbase (2.27) Lij =Sij +Sji (2.28) Finally, the Bus Power Injections result from applying the equation (2.27) in addition to the initial bus current apparent power. S∗ i= Busn X j=1 V∗ iVjYij (2.29) Where the real part is the active power Pi=real(Si) and the imaginary is the reactive power Qi=−imag(Si). Chapter 3 Validating the results 3.1 Considerations. In this chapter the process to validate the results is described and results of different cases are discussed. The process of validating consist in considering different cases and running them in a consolidated power flow algorithm. In this work MATPOWER is set as the reference. The differences between the final results in both programs are discussed with the purpose of explaining them. Once the reliability of the present work’s program is set different scenarios are tested focusing on describing the behaviour of the system. The methodology for the analysis is based on varying the value of significant parameters of the system in a range that include realistic and common situations and observe the reaction of the power flows of the system. The script contains different limitations that are commented in the next sections. Some of them are the reactive power limitation (Qmax, Qmin) when working with PV buses or the capability curve of the converters. 3.1.1 Limitations In real situations there are a lot of physical limitations that can be easily missconsidered. The aim of this section is to cite all the limitations that have been taken into account and if it is possible define a procedure to deal with each one. Capability curve of the converter In the paper [2] the dependency of the inverter’s capability curve on ambient temperature, solar irradiance, the dc voltage variation and the inverter operation. The converter used to perform the analysis found in [2] is an inverter formed by two stages; the first one is 18 Design, modeling and simulation of a PV power plant 19 a dc-dc stage used to step up the voltage to keep working on the maximum power point (MPP). The second one make the proper conversion to an ac signal. The first two parameters, ambient temperature and solar irradiance, are environmental due to it the values can not be chosen. The rest of the parameters are electrical and can be modified with robust procedures. The dc voltage variation is limited in the lower limitation (Vmin) by the minimum voltage to keep the ac side under the nominal values. These nominal values are dictated by the grid code applied. The upper limitation (Vmax) in set by the open circuit voltage of each panel times the number of panels connected in series. In the figure (3.1) the relation between the Vdc and Pdc and the main points under study are represented. Figure 3.1: Relation between the Pdc and Vdc and main points to analyze. 3.1.2 Distribution system limitations To ensure a correct and smooth penetration of the DES, especially solar energy based generation systems, many limitations are imposed. Depending on the definition of the buses, PQ or PV the remaining parameters have limitations. In first case the PQ achieve the given power values by regulating the voltage angle and magnitude. The boundaries and their strictness are set depending on the local point of the system and its possible implications on the devices connected to it. In section 4 the regulation and the examples of some grid codes can be found. In second case and analogously the P V buses maintain the voltage constant by regulating the reactive power. Thus, those are attached to the reactive power limitations that are explained in section 4 3.1.3 Parameters under study. The indicators chosen to study and validate this work’s results are listed next. •Seventh bus angle and magnitude (V7thbus): the capacitor is connected between the ground and the seventh bus therefore it is included is the analysis. Design, modeling and simulation of a PV power plant 20 •Eleventh bus angle and magnitude (V11thbus): both of the surface script and linear script impose the same changes in each of the four pv arrays (buses 11,12,13,14). Thus, the last bus included is the eleventh bus due to the balanced and symmetric grid in the present example. •Second bus angle and magnitude (V2ndbus): the second bus voltage is chosen due to the proximity to the connection point. •Total active and reactive power demand. •Total active and reactive power generation. •Real and imaginary losses. 3.2 Example case. Figure 3.2: Example case with designated buses. The figure 3.2 illustrates the system under study and the position of the buses. It remains the same for all the buses during all the scenarios. The next table 3.1 contains the value of the different elements from the system Design, modeling and simulation of a PV power plant 21 Table 3.1: Present work system parameters. Element Value Rc 8E-03 ohms Xc (4,28E-05 H) 1,345E-2 ohms Xcap1 (4,20E-04 F) 7,5788 ohms Rcab 1,475E-02 ohms Xcab (1,42E-04 H) 4,446E-2 ohms Rtr 1,60E-05 ohms Ltr (1,45E-04 H) 4,5553E-02 ohms Rtrhi 0,01 ohms Ltrhi (9,55E-05 H) 3E-2 ohms Rg 5,33E-02 ohms Lg (1,697653E-03 H) 5,33E-1 ohms 3.2.1 Per unit values The existence of transformers which have different ratio in the system make per unit values a great tool to work with. In order to change the values to per unit the base values shall be defined. The base apparent power remains the same in all the transformer sides. The base voltage is the one set by the transformer side where the impedance is located. Finally the relation between this base parameters and the current and impedance bases is set by the Ohm’s law which is represented next: Sbase =VbaseIbase =V2 base Zbase (3.1) With the base values the conversionIof all the values is performed resulting in the next table: Table 3.2: Example’s parameters in pu. Zbase1Zc(0,2 + 0,33615j) pu Zcap (-5,2778E-03j) pu Ztri (1,6E-07 + j4,5553E-04) pu Zbase2Zcab (1,475E-04 + j4,461E-04) pu Zthri (1E-04 + j1,152E-04) pu Zbase3Zg(5,333E-06 + j5,3333E-04) pu 3.2.2 Initial values and suppositions As it is discussed in the Power Flow analysis section the script need some of the parameters to start a new case. In one hand in terms of the bus data the requirements are listed in IZbase1= 0,04 ohms, Zbase2= 100 ohms and Zbase3= 10 kohms. Design, modeling and simulation of a PV power plant 22 table (3.3). Table 3.3: Required parameters for each bus. Bus Bus id Type 1Slack bus, 2PV bus and 3PQ bus. Vsp Voltage specified at the beginning for bus i. theta The V phase of bus i. PGiReal power generated at each bus. QGiReactive power generated at bus i. PLiActive power as load in bus i. QLiReactive power as load in bus i. Qmin Inferior limits for reactive power. Qmax Superior limit for reactive power. In the first cases the following suppositions are assumed: •The surface script is a method for validating the results against MATPOWER and there are no final conclusions extracted. •The first bus is always the slack bus. In the current situation the grid bus is set as bus 1 and set as angle reference. In the other hand the value of the lines is required. The parameters included are shown in table 3.4. Table 3.4: Required parameters for each bus. From bus Bus id of the beginning of the line To bus Bus id of the end of the line R [pu] Resistance of the line. X [pu] Reluctance of the line. B/2 Ground admittance X’mer TAP tap value. The suppositions for the line parameters are: •The Qmax and Qmin, in case of choosing P V buses, are set to -99 MVAr and 99 MVAr respectively that mean there are no limitations or those are considered wide enough for the systems power magnitude. •The tap valueII used to adjust the reactive power in the bus terminals is always equal to 1 in order to simplify the initial verifying operations and the existence of power electronics. IIthe tap value is a capability some transformers have to change the turns ratio. It is used to manage the reactive power. Nowadays the power electronics are more efficient and much less cost-effective. Design, modeling and simulation of a PV power plant 23 •The ground admittance referred to the admittance of the shunt elements connected to the bus. The value refers to the half of the total susceptance (B) of the node due to the value is repeated in the two branches that include the particular bus. 3.3 Result comparison versus MATPOWER With the aim of validating that the present work’s PFE script is robust enough to extract conclusions of it, different cases are tested. Two MATLAB based programs are included in this work: the surface script and the linear script. The linear script performs multiple simulations in both programs; PFE and MATPOWER changing one of the initial parameters and plots the results of the variables of interest discussed in section 3.1.3 in a 2D graphic. It is used to evaluate the effects of changing one of the inputs of the system with the aim of finding out its behaviour. The surface script performs multiple simulations changing two of the initial values and plots surfaces representing all the combinations for a particular range of the two variables. It is used to validate the results versus MATPOWER. Both of them are limited due to the impossibility of applying different conditions on each PV array. However some cases will be performed manually, applying different conditions in some of the PV arrays (buses 11, 12, 13, 14). The surface script is useful to identify interest points such as minimum or maximum values depending on the two variables changing. In the next sections a summary of the cases and the results for each program are discussed following the same structure: 1. Particular case initial values and considerations. 2. Result comparison. 3.3.1 First comparison: Active and reactive power demand variation with PQ buses. Initial values and considerations The reason for choosing the following parameters is discussed in the section 3.1.3. This case is conformed to test the capability of the PFE solver to work with PQ buses. The values used for the simulation are shown in the next table and are referenced to the buses 11, 12, 13 and 14. In the real system these buses represent the PV arrays into the collection grid: Results Firstly the results for the voltage and angle of the buses of interest are displayed in figures 3.3, 3.4, 3.5. The original colors are black with 60% of opacity and green. As the results Design, modeling and simulation of a PV power plant 30 3.4 Analysis of maximum and minimum voltage deviations and loses. In this section individual simulations and are performed with the aim of defining the behaviour of the system. For the next simulations all the converter outputs are set as PV type buses as the majority of devices need a range of voltages to work correctly and the magnitude of the changes in the system’s voltages are commonly wider than those ranges. In section 4.3 a new type of bus is defined where the reactive power is a function of the voltage with the aim of reducing the losses and therefore the stress of the system among other benefits. 3.4.1 Particular case: different values for each converter output In this case the voltage of the converters (buses 11, 12, 13 ,14) are set to different values to simulate a real situation where the differences between components and conditions set different efficiencies in each PV array. For example; different dust densities in the panels or shades that reduce the efficiency of the Maximum power point tracker (MPPT)III. The initial data for the buses is summarized in the table 3.8 Using the script Individual Table 3.8: Initial values for the converter output buses. Bus id Voltage [p.u.] P demand [MW] P generation [MW] 11 1.02 0.65 0.43 12 1.01 0.59 0.44 13 1.03 0.62 0.41 14 1.09 0.66 0.40 case the results are obtained and represented in figure 3.12. It can be observed that the maximum voltage deviation in per unit, excepting the voltage regulated buses, is found in bus 5 with an increment of 0,0161% which transformed to real values is an increment of 322 V. The minimum voltage deviation is found in bus 2 with a per unit increment of 0,0094% and real value of 188 V. The losses and injected apparent powers in each bus are illustrated in figure 3.13. The real and imaginary losses are similar to the magnitudes of thesurface script for the same grid voltage value and active power generation figure 6.10. In fact if the linear script is used to evaluate the relation between the grid voltage and both imaginary and real losses using the same initial data. The results from figure 3.14 correspond to a variation in grid voltage from 0.9 to 1.1 p.u. It can be concluded that both losses present a minimum when the grid voltage is 1. IIIMPPT: is a device that set the voltage in the array to meet the maximum active power output. It has to work with the lower voltage on the array therefore if one panel is performing poorly the rest of the array is affected. It is considered as a type of loss. Design, modeling and simulation of a PV power plant 31 Figure 3.12: Example scheme with voltage magnitude and phase values Figure 3.13: Example scheme with voltage magnitude and phase values Design, modeling and simulation of a PV power plant 32 Figure 3.14: Real and Imaginary losses changing the grid voltage from 0.9 to 1.1 Chapter 4 Grid code requirements 4.1 Introduction A Grid Code is a technical document containing the rules governing the operation, maintenance and development of the transmission system. It is necessary due to the existence of different energy suppliers that work with the same Transmission system operator (TSO). In the grid code all the requirements are specified and discussed to ensure stability and quality of electrical systems in continuous operation as well as during fault’s or extraordinary events. Grid codes are different and depend on the country they are conformed. Each country elaborates one and the differences between them are motivated by the characteristics of the country and the particular needs created by the local electrical system. However according to [7] there is a tendency to harmonize grid codes in different EU countries with documents such as: Network Code on Requirements for Grid Connection applicable to all Generators (RfG) drafted by European Network for Transmission System Operators for Electricity (ENSTO-E). These documents are used as framework while developing grid codes. The increasing penetration of renewable energies during the last decades have forced improvements in the technology and system’s behaviour control. These changes are mainly motivated by the intrinsic characteristics of renewable energies, such as the dependency on the environment and the intermittent behaviour of the natural energy resources. These variations can conclude in undesired faults creating damages to the transmission system elements and all the equipment connected to the network. The grid codes are constantly updated to ensure smooth and correct functioning in all situations. For example the grid code of a large country with a good number of sources and a low penetration of renewable energies must have less strict conditions for transmission system frequency than a little island which has an equal proportion of renewable and non-renewable energy sources. There are two types of requirements: static and dynamic. Static requirements refer to the continuous operation state and the dynamic collect the procedures to follow during fault sequences and disturbances. The grid codes include worst case scenarios to cover 33 Design, modeling and simulation of a PV power plant 34 all the possible events that can take place including strict measures as the complete disconnection. The renewable energies present a big flexibility to penetrate in the market as they can be introduced as wind farms, PV plants, Hydro-powered stations... The solar energy is one of the resources that are experiencing a fast growth during the last decades. In the paper [6] presented in MPDI journal it is exposed that the power installed capacity in Spain for solar energy increased over a 5 GW from 2007 to 2018 with a positive tendency. The evolution is represented in figure 4.1. Figure 4.1: Power installed in Spain from 2007 to 2018. In the present work the grid code requirements for a Large Scale PV power plant (LSPVPP) are discussed and contextualized for different grid codes. The particular solution presented in [8] for compensating reactive power is introduced in the present work PFE solver. According to [11] the amount of energy managed via the peninsular system’s ancillary services in GWh is illustrated in figure 4.2. Figure 4.2: Summary of the total energy managed by the ancillary services in the spanish peninsula. Design, modeling and simulation of a PV power plant 35 4.2 Static regulation In general the static regulation for renewable energies depend on the resource used and its intrinsic characteristics. For example for solar and wind sources power curtailment is needed due to the unpredictable availability and the fast and wide variations they can suffer due to environmental conditions. Power curtailment limits the power extraction in two different situations: when the power extracted is greater than the nominal value of the plant in peak generation events or when the demand is lower than the production. In [4] an analysis of the connection requirements of wind power generation units for six grid codes is carried out. 4.2.1 Power factor regulation The power factor is the ratio of reactive to apparent power consumption, thus its requirements and the reactive power are similar in the different Grid Codes. Any inductance with no capacitors attached consumes reactive power. Consequently this reactive power consumption must be produced somewhere in the grid. One thing to consider is that the distribution of reactive power is cost intensive and the reactive power is not profitable although it is used in some cases to support the grid during particular events. In [5] an example of different Grid Codes power factor/reactive power requirements is presented: Figure 4.3: Reactive power requirements for Puerto Rico, South Africa and China, Germany and Romania. The most restrictive PF requirement in figure (4.3) is the red line referring the Grid code from South Africa. From 100% to 20% of production the maximum PF are 0.975 leading and 0.975 lagging. One of the limitations commented in section 3.1.1 are the reactive power bounds for PV buses. Those limitations should be established in order to meet the Grid Code guidelines Design, modeling and simulation of a PV power plant 36 of reactive power consumption. 4.2.2 Power Curtailment Power curtailment introduces the possible reduction of generated active power depending on grid requirements. Two common situations where power curtailment is effective are: •The possible overloading at peak generation hours for natural resources as wind or solar energy. Due to the inherent characteristics of those. •When the demand is lower than the generated active power without power curtailment. The overloading is regulated by the TSO because it would cause an increment of the frequency, therefore an increment in the inertia of the transmission system. The power curtailment is not outlined precisely in most of the Grid Codes and it depends on environmental conditions, installed capacity and economics. According to [4] an indicator to apply this requirement is the frequency surplus. In European countries, with a grid frequency of 50 Hz, the bound to start applying power curtailment was 50,5 Hz. When the frequency deviation decreases the reestablishment of the active power has its own limitations. It should be less than the 10% of the network capacity per minute. Figure 4.4 extracted from [4] illustrates the operation conditions when the network is overloading. Figure 4.4: Power curtailment requirements 4.2.3 Active power reserves The tendency of introducing more and more Distributed energy sistems (DES) based on renewable energies in existing distribution networks generates new problems. The difficulty of prediction and intermittency of renewable energy sources in addition to the high density of DES lead to voltage and frequency issues in the distribution network. For example the proximity of PV plants due to the land restrictions submit them all to the same environmental conditions. It has to be noticed that the most common peaks of solar production are found in the middle of the day where the demand is low. The energy surplus is redirected into the grid however in a distribution network with a high density Design, modeling and simulation of a PV power plant 37 of DES it can conclude in bidirectional power flows, overloads or over voltages on some branches or transformers. To fix this issues and ensure a smooth penetration of the DES in the distribution networks two solutions are found. The first one is the use of batteries as energy storage systems to supply the active power demand in compromised situations. Although it has been experimentally implemented presenting good results it is still cost-prohibitive and is not cost-competitive if compared with other sources. The main problem is the current energy storage technology that presents a low efficiency and a high maintenance and production costs at this power ratings. The alternative method is the active power regulation. It is based on the power curtailment. In a situation where the energy production can reach the nominal values the active power output is regulated to offer around an percentage of the maximum capacity. The remaining capacity is set as active power reserves. This approach is easier to implement but it causes a reduction in the efficiency of the plant. 4.2.4 Voltage range and Control The PV plant must be able to run at rated voltage plus the specified voltage range. The rated voltage and the voltage range varies depending on the transmission system, which depends on the country. The voltage ranges at the different rated values, for limited periods of time during fault or particular events, are also specified. According to [4] the requirement for all nominal values in continuous operation varies around ±10% of the nominal value. 4.2.5 Remote Voltage Control Additionally in most of the grid codes it is required that the LS-PVPP or power station contains a closed loop voltage regulation system. The voltage regulation follow the basis of the PV bus definition where a voltage set point is specified and regulated by continuously modulating the reactive power output in the acceptable ranges. In addition the voltage set point shall be within the limits. The time response of the voltage regulation system is also specified in some of the grid codes analyzed in [4]. 4.2.6 Frequency The generator’s plant must be capable of working in accordance with specified frequency ranges. The majority of the grid codes presented [4] must be able to run continuous operation in frequencies between 47,5 Hz and 52 Hz. Time delays might be necessary in some range of frequencies. The procedure in case of breaking the limits is to reduce the active power output and remain connected. Disconnection is only permitted at frequencies below 47 Hz and above 53 Hz and with no time delay. Design, modeling and simulation of a PV power plant 38 4.2.7 Flicker A flicker is a single, rapid change of the root mean square voltage. The connection of different elements such as capacitors, lines, cables, transformers and other elements may cause transmission system step changes. It must be mentioned that a lot of grid codes do not mention flicker’s, according to [4] the Danish grid code specifies the next procedure. Different thresholds are set depending on the time interval where the flicker magnitude is evaluated. The short term flicker is a weighted average of the flicker contribution over ten minutes. The long term flicker is defined over 2 hour. Finally in figure 4.5, extracted from [5], a summary of the requirements for a LS-PVPP is illustrated for a better understanding. Figure 4.5: Summary of grid codes requirements for LS-PVPP 4.3 Grid support; reactive power compensation As the present work is centered in renewable energies and they are based on natural resources usually the generation plants are located in remote areas. Consequently the transmission lengths are long enough to consider the transmission line impedance significant. As the transmission lines are mostly composed of inductance elements, which are part of the imaginary domain, and transformers. The inverter is synchronized to a point of connection (PoC) and the feedback measurements are taken at this point. As the current fed to the PoC flows through the grid impedance a quantity of reactive power is created although the inverter is working at unity PF. Thus, the active power of the PV inverter becomes coupled with the reactive power seen by the grid. This undesired reactive power causes a bad behaviour that includes an increase in transmission losses and consequently reduces the maximum transmission capacity, compromises system stability and strains the grid. The undesired reactive power depends on the grid current and grid impedance. If the impedance is high the PoC current has a significant impact on the local voltages. A grid Design, modeling and simulation of a PV power plant 39 Figure 4.6: Scheme of the connection points contextualized in the current example with a high impedance is defined as a weak grid. A parameter to measure the degree of weakness is the Short circuit ratio (SCR) which is defined by the next expression: SCR =V2 n SnZg (4.1) Where Vnis the nominal voltage, Snis the nominal apparent power (usually in distribution generation systems is the value of the inverter due to the transformers have the same power ratio) and Zgis the transmission system impedance. Once the value of the SCR is less than 5 it can be considered weak. As it is commented in previous sections the renewable energy generating plants are located at remote locations the impedance become significant, thus the grid is considered weak. According to [8] the reactive power has been managed using synchronous generators, capacitor banks as the present work case examples or with FACTSIdevices. The reactive power of an inverter is relatively easy to control without the need of additional investments. The reactive power control has two goals: voltage regulation and reactive power optimization or a compromise between them. In fact they can not be optimized simultaneously as they are related. In [8] and in the present work the voltage regulation is not considered as the buses are set as PV buses and a notification pops up if the Q limits are violated with the current voltage setpoint. According to the authors of [8] the most common approach to control reactive power in LS-PVPP is the Power factor control (PFC) where the reactive power is proportional to the active power feed. Conventional PFC assumes the active power-reactive power ratio (Q/P) constant although the reactive power injection from inverter affects the PoC voltage changing the commented Q/P ratio and therefore the PFC achieves optimal decoupling only when the inverter is operating at nominal values. In renewable generation systems the active power output changes drastically along with the environmental conditions therefore if traditional PFC is applied it leads to suboptimal reactive power compensation. IFACTS: flexible alternating current transmission system: system of static components used for the AC transmission used to enhance control and increase power transfer capability for the network. It is usually based on power electronics. Design, modeling and simulation of a PV power plant 46 deep study on the location, a mitigation plan and a correct law code that protects the more sensitive areas can conclude in many environmental and social benefits. Chapter 6 Projects budget The present work is centered in the optimization of a PV plant by including a tool that allows the user to take advantage of real time information to reduce losses and choose the best settings at each situation. The cost of implementation is relatively low in comparison to the total budget of creating an entire PV power plant and it can have positive economic effects as consequences. Different conception costs per energy unit are presented. In first place the CAPEXI(capital expenditure, table 6.1) data for a LSPVPP project are presented. In second place OPEXII (Operational expenditures, table 6.2) derived from the continuous operation of the PV plant are tabulated. Finally a small review of the economic impact of the ancillary services is presented according to Red El´ectrica de Espa˜na. 6.1 Large Scale PV power plant The prices shown on table 6.1 are extracted from [2]. The data is extracted from Bloomberg New Energy Finance which is a reliable energy related database. The battery system commented in section 4 is included to illustrate the high cost per kWh of this methodology. The miscellaneous cost include an average land prince and the engineering, the derived cost of legal permissions and the construction among others. Where EUR/Wnis the cost per watt of nominal power. The Balance of the system (BOS) encompasses all the components of a photovoltaic system other than photovoltaic panels and the inverters. This concept includes wiring, switches, mounting system and some mechanism used to increase the efficiency of the PV power plant such as the maximum power point tracker (MPPT) or the GPS solar tracker that changes the orientation of the panels depending on the position of the Sun. In the table 6.2 the continuous operations cost for a year of production is presented. The PV technologies need energy to start converting energy and the intermittency of the solar ICAPEX; the inversion cost to start running the PV plant. IIOPEX; the operation and maintenance cost of running the PV plant. 47 Design, modeling and simulation of a PV power plant 48 Table 6.1: PV plant CAPEX costs and current project budget CAPEX component Cost per Watt Current project Modules 0,31 EUR/Wn1,24 M. EUR Central inverters 0,03 EUR/WAC 0,12 M. EUR Balance of the system 0,17 EUR/WAC 0,68 M. EUR Miscellaneous cost 0,30 EUR/WAC 1,2 M. EUR Power curtailment with batteries (optional) 400 EUR/kWh - source or the night periods create situations where the consumed energy is higher than the production. The dust removal, panel cleaning, maintenance products and salaries are included. The tracker system operation cost is isolated to illustrate the associated cost of implementing a solar tracking system. Table 6.2: PV plant OPEX OPEX component Cost per MW OPEX PV plant operation 9500 EUR/MW OPEX tracker operation (if included) 900 EUR/MW 6.2 Ancillary services The grid codes requirements have an impact on the average final price of energy. In the report of Red El´ectrica de Espa˜na of 2018 about the ancillary services [11] a breakdown of the components conforming that price is performed. The figure 6.1 illustrates it. The grid support requirements seen in section 4 introduce a reduction on the final price. The magnitude of its reduction is small in comparison with the final price but the advantages in energy management in terms of reliability and security make their application essential. Design, modeling and simulation of a PV power plant 49 Figure 6.1: Breakdown of ancillary services cost in the average final price of energy in the peninsular system Conclusions and further work During the last years solar energy has penetrated in the energy market as an alternative to fossil fuel-based energy. It has been possible due to the advances in power electronics mixed with the automatic control methodologies that permitted the mitigation of the unintended behaviour of renewable energies. The balance of the environmental effects brought by these renewable energies is mostly positive however in section 5 is shown that the effects on the environment are diverse and strongly dependent on the situation and the ecosystem sensibility. In some cases those consequences are unknown and can conclude in a permanent change in the surroundings of the installation with important losses in biodiversity and natural wealth. In consequence a deep study on the biodiversity and present ecosystem shall be performed in order to mitigate those negative effects. The soil and physic environment are also affected as a part of the energy they were usually receiving to develop natural processes is used to produce energy for human consumption causing changes in the local temperature and soil nutrients development. In the current work a program based on [3] to solve the power flows of a given system is developed and validated versus MATPOWER. It models the buses using traditional methods as PQ and PV buses. A summary of the most common static regulation present in most of the grid codes is included. A particular solution to mitigate the negative effects of the unintended reactive power on large distribution systems based on regulating the amount of reactive power as a function of the voltage is also included. In further work the implementation of that bus model and more grid code requirements can be performed. Although the good results achieved by energy storage systems and the new developments the implementation cost is still high in order to be competitive. Thus, the active power reserves methodology discussed in section 4.2.3 leads to a efficiency decrease. 50 Annex 6.3 Introduction In this chapter different images and script information from this work can be found. 6.4 Scripts A brief description of the scripts used in the present project is included next. The scripts can be found in the .zip file annexed to this project. •The scripts named busdatas and linedatas are extracted from [3] and are used to introduce each case initial data. The information about the data introduction is discussed in section 3.2.2. •ymatrix is used to create the admittance matrix. •pfeq and pfeq2 are the core of the NRA. Their results are the voltage and phase values of the buses. The difference between them is the way of displaying the results. •loadflows1 and loadflows2 are used to perform the load flow analysis. They are called from pfeq and pfeq2 respectively. •The Surface script is used to validate the results against MATPOWER and identify particular points if needed. It varies two parameters of the system and plot the results in different surfaces for both programs. The different parameters that can be changed are: active and reactive power demand (Pl,Ql), active and reactive power generation (Pg,Qg), all converters voltage at the same rate (only PV buses; V11,V12,V13 and V14) and the grid voltage (V1). •The linear script is made for a two dimension comparison to study the behaviour in a more isolated manner than in the Surface script. •The linear pfe script perform simulations of changing one parameter and plots the results in two dimension graphics. The available parameters to change are the same from surface script. 51 Design, modeling and simulation of a PV power plant 52 •The individual case script perform one simulation of the selected case and displays an accurate summary of the values of the system. A breakdown of the branches and buses values as well as the totals are included. In addition in cases including PV buses the limit violations are also displayed. 6.5 First comparison power flow results (a) Active power demand in KW (b) Reactive power demand in KVAr Figure 6.2: Active and reactive power demand (1st case) (a) Active power generation in KW (b) Reactive power generation in KVAr Figure 6.3: Active and reactive power generation (1st case) Design, modeling and simulation of a PV power plant 53 (a) Real losses in KW (b) Imaginary losses in kVAr Figure 6.4: Imaginary and real losses (1st case) Design, modeling and simulation of a PV power plant 54 6.6 Second comparison power flow results (a) Active power demand in KW (b) Reacive p`ower demand in KVAr Figure 6.5: Active and reactive power demand (2nd case) (a) Active power generation in KW (b) Reacive p`ower generation in KVAr Figure 6.6: Active and reactive power generation (2nd case) (a) Real losses (b) Imaginary losses in kVAr Figure 6.7: Active and reactive power generation (2nd case) Design, modeling and simulation of a PV power plant 55 6.7 Third comparison power flow results (a) Active power demand in KW (b) Reactive power demand in KVAr Figure 6.8: Active and reactive power demand (3rd case) (a) Active power generation in KW (b) Reactive power generation in KVAr Figure 6.9: Active and reactive power generation (3rd case) (a) Real losses (b) Imaginary losses Figure 6.10: Active and reactive power generation (3rd case)