Performance Analysis of Naàma’s 20 MW Grid-Connected Plant in Semi-Arid Climate in Algeria

Abstract

This article is devoted to the study of a 20 MW large-scale photovoltaic power plant (LS-PVPP), connected to the grid and located in Naàma, Algeria. The power plant is included in the National Program for the Development of Renewable Energies 2015–2030. Among the parameters analyzed in detail in this work, the performance ratio recorded an average value of 67.55%, the capacity factor had an average of 17.10%, the total losses had an average of 2.10 kWh/kWp/day, the system efficiency had an average of 4.10 kWh/kWp/day and an annual average of 9.84% of the efficiency. A linear regression equation with a coefficient of determination R² of 0.91 confirms the importance of irradiation impact in the region; less significant linearity for the effect of temperature with a coefficient of determination R² = 0.28 is recorded for production. A comparative study conducted with the Adrar plant (Algeria) with an extremely hot desert climate and the Saida plant (Algeria) with a semi-arid climate demonstrated that the efficiency of the Naàma station is equal to 91.22% of the efficiency of Adrar and 73.47% of the efficiency of Saida. Naàma is known for its semi-arid climate; it is very cold in winter and hot in summer, with sandstorms becoming more frequent due to climate change. PVsyst software (Version 7.4.8) is used to validate the results.

1. Introduction

Algeria is located in a geographical region that offers one of the highest sunshine potentials on earth. It was only since 2011 that this advantage has attracted interest in political circles [1] when the government adopted the National Program for the Development of Renewable Energies, which consists of installing power of 22,000 MW by 2030 for the national market (Figure 1). Overall, the electricity production park consists of the power plants of the Algerian Electricity Production Company (SPE), which run on natural gas, and the photovoltaic and wind power plants of Shariket Kahraba wa Taket Moutadjadida (SKTM) [2]. The shares of photovoltaic power plants are, respectively, 3000 MW for the first phase (2015–2020) and 10,575 MW for the second phase (2021–2030) [3]. The first pilot photovoltaic plant was installed in 2014 in Oued N’Chou in the Ghardaia province with an arid desert climate, and it injects 1.1 MW into the network. Four technologies (monocrystalline, polycrystalline, amorphous and thin-film CdTe) are mounted on fixed and motorized structures. But no research work has been conducted on this station and no comparison has been made between these technologies in order to better choose the type of cells for arid climates.

Up to February 2017, SKTM produced 343 MW of photovoltaic energy, and in the next step, the development of an electrical interconnection between the North and the Sahara (Adrar) will allow the installation of large renewable energy plants in the regions of In Salah, Adrar, Timimoune and Bechar and their integration into the national energy system. It is in this context that several large-scale photovoltaic power plants have been installed across the country. The choice of sites is based on their proximity to small towns in the south and highlands.

Many studies have been conducted and published in the field of the performance evaluation of LS PV systems. These studies can be categorized according to certain characteristics. The installation can be a floating type [4] or rooftop [5,6,7]. It can generate small power [8] or large power [9]. It can be connected to the grid [10] or exist a standalone system [11]. It can also be placed in a cold [12] or tropical [13] region, a Mediterranean climate [14] or arid and hot climates [15,16]. Among the evaluation criteria, there is also the technology used in photovoltaic cells such as monocrystalline silicon, polycrystalline silicon or thin-film modules [17].

The Naàma photovoltaic plant is a part of the national program for the development of renewable energies. It produces 20 MW that it injects into a 60 kV medium-voltage network. This article is dedicated to the performance evaluation of this plant and also includes a comparison made on the basis of data published in research articles with two other plants in the same program. The first plant is located in Adrar and injects 20 MW into the autonomous electricity network PIAT (Pole In Salah, Adrar, Timimoun), and the second is located in Saida and injects 30 MW into the RIN (National Interconnected Network). The plant managers provided us with data for the year 2017 concerning the energy produced, the energy injected into the network and the meteorological data. This allowed us, over the months, to evaluate the variations in the performance ratio, the capacity factor, the different yields, losses and efficiencies.

2. System and Site Description

2.1. Weather Profile for Installation Site

Sedret leghzel is the exact location of the 20 MW Naàma power plant. It is located 5 km west of the city, on open terrain, without relief and without vegetation (Figure 2). This leaves the plant unprotected against the gusts of sand winds that come from the east and that are increasingly frequent due to climate change. The site is at an altitude of 1105 m, with a latitude of 33.5° N and longitude of −0.83° W, with a semi-arid climate.

The percentage of cloud cover in Naàma experiences considerable seasonal variation throughout the year. The clearest period in Naàma lasts for 3 months in the summer. The cloudiest period lasts for 9 months, with cloud cover occurring 39% of the time. Thus, the amount of solar energy received by the site remains significant throughout the year.

The temperature varies between an average of 8.5 °C in a harsh winter with frost and snow and an average of 30.1 °C in a hot and dry summer.

2.2. A Description of the Power Plant

Naàma’s power plant is connected to the RIN grid on the 60 kV busbar; this part of the grid is supervized by the Electricity Transmission Network Management Company (GRTE). The station contains 79,680 CS6P-250PX polycrystalline silicon modules (Canadian Solar Inc., Guelph, ON, Canada), 20 SMA SC800CP-XT inverters (SMA Solar Technology AG, Niestetal, Germany) and 10 SGB STARKSTROM (1800 kVA/30 kV) transformers (SGB-SMIT Group, Regensburg, Germany). These components are contrasted with the components of two other sites, Adrar and Saida, for a comparative study (

Table 1. Comparison table of characteristics.

Design Parameters	Characteristics
Site	NAAMA	ADRAR	SAIDA
Latitude	33.5° N	27° N	34° N
Elevation	1100 m	279 m	900 m
Array rating	20 MW	20 MW	30 MW
Orientation and tilt	15°, S	26.5°, S	15°, S
Installation type	Fixed	Fixed	Fixed
Distance between rows	6 m	8 m	-
PV technology	p-c Silicon, CS6P-250P	Yingly Solar YL245-29b	p-c Silicon, CS6P-250Wp
Number of modules	79,680	81,840	119,520
Inverter power	800 kW	500 kW	954 kW
Number of inverters	20	40	30
Type of transformers	1800 kVA/900 kVA/900 kVA	1250 kVA/630 kVA	-
Number of transformers	10	20	15

).

The AC electrical network is organized into two loops; each one contains five identical subfields (called skids). Each of the skids is composed of two parallel inverters and one transformer. Each of the loops collects the power of five transformers via five ring main units (RMUs). Modules are assembled with 24 per string. The entire system is controlled and the acquired data are registered in a data unit, which is placed in a control room.

3. Performance Parameters

Environmental parameters strongly impact the production of the electrical energy of the PV system. In order to analyze this impact, it is necessary to calculate (according to IEC 61724 standards [18]) the performance parameters using mathematical models of the different parameters that are available in most of the articles that deal with power plants similar to the Naama plant. The main parameters are listed below.

3.1. Yields

3.1.1. Reference Yield (Y_R)

The reference yield is assessed as the ratio between the total solar radiation incident on the surface of the PV solar panels H_t (kWh/m²) and the reference radiation quantity G₀ (1 kW/m²).

$$Y_R={\displaystyle \frac{H_t}{G_0}} \text{(kWh/Kw/day)}$$

(1)

G₀ is the reference irradiance at STC (1000 W/m²).

3.1.2. Final Yield (Y_F)

The final yield is calculated as the ratio between the total energy produced by the system Eac (kWh) and the installed nominal power P₀ (kW).

$$Y_F={\displaystyle \frac{E_{ac}}{P_0}} \text{(kWh/Kw/day)}$$

(2)

3.2. Performance Ratio (PR)

The performance ratio indicates the overall effect of the losses on the energy production of the PV system. It is defined as the ratio between the final yield, Y_f, and the reference yield, Yr.

$$PR={\displaystyle \frac{Y_F}{Y_R}} \text{(\%)}$$

(3)

3.3. Capacity Factor (CF)

The capacity factor is calculated as the ratio of actual electrical energy output over a given period to the theoretical maximum electrical energy output over that period.

$$CF={\displaystyle \frac{\frac{E_{ac}}{P_{MPPrated}}}{24\times numberofdayinmonth}}\times 100\left(\%\right)$$

(4)

3.4. Efficiency

3.4.1. PV Module Efficiency (η_pv)

PV module efficiency is calculated as the ratio between Edc power and the solar energy captured. It can be calculated per day or month.

$${\eta }_{pv}={\displaystyle \frac{E_{dc}}{H\times A_m}}\times 100\left(\%\right)$$

(5)

H is the solar energy received by the PV system, and A_m is its area.

3.4.2. Inverter Efficiency (η_inv)

The inverter efficiency is assessed as the ratio between DC power Edc of the system and energy output Eac.

$${\eta }_{in}={\displaystyle \frac{E_{dc}}{E_{ac}}}\times 100(\%)$$

(6)

3.4.3. System Efficiency (η_sys)

$${\eta }_{pv}={\displaystyle \frac{E_{ac}}{H\times A_m}}\times 100 (\%)$$

(7)

3.5. Losses

Some phenomena can cause DC side losses such as shading, fouling and component failure. On the other hand, AC side losses are due to a decrease in the efficiency of transformers and inverters.

3.5.1. Array Losses (L_C)

Array losses represent all the losses that occur during the operation of the PV modules:

L_C = Y_R − Y_A (kWh/kW/day)

(8)

3.5.2. System Losses (L_S)

System losses represent inverter losses and are the difference between the array yield and final yield as follows:

L_S = Y_A − Y_F (kWh/kW/day)

(9)

4. Results and Discussion

This study of the 20 MW large-scale PV power plant situated in Naàma focuses on the solar potential assessment of the site. It is based on the monitored collected data via SCADA from 1 January to 31 December during the year 2017 and on the simulated values with PVsyst software.

4.1. Meteorological Data

The meteonorm 8.1, 2021 database used by PVsyst 7.4 collects its data from weather stations worldwide. After verification, the location of Naàma and the two locations of Saida and Adrar, which will be used for comparison, are shown on the layout map of the 8325 meteonorm stations [19].

Table 2. Monitored and simulated weather-dependent data for 2017.

	Monitored					Simulated
Month	Temp °C	Win m/s	Re hu %	GloInc kWh/m2	Energy MWh	Temp °C	Win m/s	Re hu %	GloInc kWh/m2	Energy MWh
Jan	6.03	6.34	62.48	133.27	1556.39	7.70	2.30	67.30	136.10	2051.10
Fab	11.99	7.10	48.10	145.47	1930.11	9.70	2.80	50.80	144.00	2134.90
Mar	14.14	5.87	40.51	204.29	2324.99	14.30	3.10	44.90	192.50	2754.70
Apr	18.64	5.53	35.59	213.58	2442.97	18.10	3.40	37.30	214.20	2984.20
May	24.95	6.36	31.01	225.43	2436.97	22.80	3.30	32.40	221.80	3010.30
Jun	28.34	5.31	26.43	225.04	2469.79	27.80	2.90	28.90	230.20	3038.40
Jul	31.97	7.13	16.91	231.38	2426.28	32.40	2.80	15.50	233.80	2982.70
Aug	30.01	11.61	26.58	203.35	2147.60	30.70	2.50	29.20	224.30	2909.10
Sep	25.34	4.79	29.81	202.38	2277.91	25.20	2.50	37.20	191.20	2583.70
Oct	19.02	3.82	35.27	186.72	2154.26	20.30	2.20	36.90	175.90	2466.50
Nov	13.24	4.42	33.46	143.12	1725.61	12.70	2.30	40.60	146.70	2149.90
Dec	7.49	5.59	60.57	123.56	1478.98	8.80	2.10	61.80	129.70	1946.80
Year	19.26	6.15	37.23	2237.59	25,371.86	19.22	2.68	40.20	2240.4	31,012.4

presents monitored and simulated data for the year 2017 for weather-dependent parameters such as ambient temperature, wind speed, relative humidity, solar irradiation and energy injected into the grid.

4.2. Temperature

With a harsh winter and a fairly warm summer, the measured monthly average ambient temperatures vary between 6.03 °C in January and 31.97 °C in July. However, the simulated values vary between 7.68 °C in January and 32.42 °C in July. Figure 3 shows that the measured values and simulated ones with PVsyst 7.4 are very close.

In the absence of real data on the module temperature, several studies have been conducted to determine it, based on meteorological parameters such as ambient temperature, wind speed and solar irradiance. Three models are chosen for the simulation and to make a comparison with the monitored ambient temperature. The first is that of Lasnier and Ang [20,21], specific for polycrystalline silicon photovoltaic modules:

$$T_{m1}=30+0.0175\left(G-300\right)+(T_a-25)$$

(10)

In the second model and according to Muzathik [22], the statistical results of the model record a total error of less than 3% of the expected temperature. The following formula considers ambient temperature Ta (°C), wind speed Ws (m/s) and solar radiation G (W/m²):

$$T_{m2}=0.943T_a+0.0195G-1.528Ws+0.3729$$

(11)

In the third model presented by Ihaddadene [23], the module temperature depends only on the ambient temperature; thus, the monthly variations in Tm and Ta occur simultaneously with close gaps. The relationship between these two parameters is linear and given as follows:

$$T_{m3}=1.032T_a+5.206$$

(12)

The measured values of the solar radiation and ambient temperature of 18/12/2018 are presented with the module temperature, simulated with the three models T_m1, T_m2 and T_m3. Figure 4 shows the three different shapes for Tm. For negative ambient temperatures, the module temperature is lower than the ambient temperature for the first two models. For low positive temperatures, Tm moves away from T_a and then begins to follow the shape of G by widening the gap with T_a. At 14 h: 15 mn, a decrease in illumination (probably due to a cloud) caused a decrease in Tm, explained by the strong dependence of T_m on G for these two models.

4.3. Wind

Current climate change has affected wind by causing a significant increase in wind speed. With a yearly average of 5.76 m/s, the monitored monthly evolution ranges from 3.82 m/s in October to 7.13 m/s in July, while it ranges from 2.1 m/s in December to 3.4 m/s in April for simulated variations, with a yearly average of 2.68 m/s.

The monthly evolution of the average daily monitored wind speed and simulated wind speed presented in Figure 5 do not match as much as in temperature and irradiation. The measured values can be considered more reliable since the Algerian highlands experience regular gusts of wind, especially during the changing seasons. These winds lift the sand and considerably affect the efficiency of the installation due to the decrease in solar illumination.

4.4. Relative Humidity

Relative humidity in the Naàma site is one of the main parameters. Its high rate allows residues, carried by very frequent sand winds, to stick to the PV modules throughout the year. Figure 6 shows strong similarity between the measured and simulated values [24,25,26].

4.5. Daily Solar Insolation

Daily solar insolation is the amount of solar energy received on a unit surface over a day and is expressed in kWh/m²/d. The monitored monthly average insolation as seen in Figure 7 ranges from 3.98 kWh/m²/d in December to 7.50 kWh/m²/day in June, while the simulated monthly average insolation ranges from 4.18 kWh/m²/day in December to 7.67 kWh/m²/day in June.

4.6. Weather-Dependent Parameters

Table 1 recapitulates the monitored and simulated monthly average daily variations in climate parameters already detailed above. A considerable yearly value of 2237.59 kWh/m² is monitored for the plane of array irradiance (POA, which corresponds to the GlobInc value in the simulation). It is very close to the simulated value of 2240.40 kWh/m². The monthly average variation in irradiation ranges from 123.56 kWh/m² in December to 231.38 kWh/m² in July, while the simulated variations range from 129.70 kWh/m² in December to 233.80 kWh/m² in July. The results are presented in Figure 8.

The monthly average daily variations in energy produced by the PV plant, the energy injected to the grid and their respective simulated values are presented in Figure 9. The annual average daily variations in energy produced, 26,195.96 MWh, and energy injected into the network, 25,371.86 MWh, have very close values, with a slightly higher value for the energy produced. The same goes for the simulated produced value, 31,479.73 MWh, and the simulated injected value, 31,012.37 MWh. The monthly average daily variations in energy ranges from 1528.62 MWh in December to 2555.50 MWh in June for the produced energy and ranges from 1478.98 MWh in December to 2469.79 MWh in June for the injected energy.

Figure 10 shows the daily variation in energy produced on 18 December as a function of the daily variation in irradiance. Figure 11 shows the daily variation in energy produced on 25 July. When comparing the two figures, it turns out that the linearity of the relationship between irradiance and energy produced is proven, both in winter and in summer. The daily average temperature was 2.98 °C on 18 December and 30.15 °C on 25 July. This proves that the energy in the studied site is proportional to the irradiance.

Figure 12 illustrates the positive effect of temperature and irradiation on the energy injected into the grid. Figure 12a shows a linear regression equation with the coefficient of determination R² equal to 0.9132, confirming the importance of the irradiation impact on the electric energy production in the Naàma region. Figure 12b shows less significant linearity for the effect of temperature on the energy injected into the grid, with a coefficient of determination R² = 0.2759 of the linear regression equation.

5. Performance Analysis

Energy losses in PV generators have multiple causes such as thermal losses, shading, and soiling. For these reasons, the energy injected to the grid never reaches the ideal value equivalent to complete efficiency. The ratio of this transferred energy is defined as the performance ratio PR. The monthly average daily PR and temperature, monitored in Naàma LS-PPPV during 2017, are shown in Figure 13, highlighting the inverse proportionality between them. The PR ranges from 61.68% in July to 78.05% in February. With an average value of 67.55%, it is less than the simulated PR estimated as 82.25%.

A theoretical and experimental study on the effect of Saharan sand dust on solar panels is published in [27]. This effect was found to be negative since, according to this study, the efficiency and performance ratio decreased with an increase in fouling. Figure 13 of this article illustrates the PR decrease rate as a function of the fouling rate, where the simulation results for fixed PV systems indicate that, in July, PR could have decreased from 73.53% to 70.43% and 63.98% under fouling rates of 0% to 5% and 15%, respectively.

If a projection of this study is made on the Naàma station, which is located in a dust-prone region, which contains 79,000 photovoltaic modules, the cleaning protocol implemented in the station is insufficient. The region also has a high humidity rate as already mentioned above, which promotes the deposition of dust on the surfaces of the PV modules. This may explain the difference between the measured rate of the PR and the simulated one.

According to the database for 2012 of the station located in the desert of Oman, producing 1.4 kW and connected to a network, the performance ratio evolved from 0.52 in July to 0.72 in December, with an annual average of 0.65 [28].

In the literature, the inverse proportional relationship between performance ratio PR and temperature T is not always proven. According to [29], during 2019, in an 81.90 kWp PV system installed on the roof of academic buildings, the PR varied in a non-regular way between 0.65 and 0.77, independently of the shape of variation in temperature, which increased from 14.46 °C in January to 34.76 °C in May and then decreased to 16.30 °C in December.

The reference system efficiency (Y_R) of the power plant was measured throughout 2017, with extreme values showing significant differences being recorded. The maximum Y_R is measured in June with the value of 7.5 kWh/kW/day against a simulated value of 7.67 kWh/kW/day, while the minimum measured value is 3.98 kWh/kW/day in December against a simulated value of 4.18 kWh/kWp/day. It is worth adding that the annual averages of the measured and simulated Y_R are, respectively, 6.12 kWh/kW/day and 6.13 kWh/kW/day.

The final yield (Y_F), defined as the array daily output energy, is also measured and compared to the simulated values. For the same period from 1 January to 31 December 2017, the month of June recorded the highest value of 4.84 kWh/kW/day against the 5.96 kWh/kW/day simulated value, while the minimum value was recorded in December with 2.81 kWh/kW/day against the simulated value of 3.69 kWh/kW/day. The annual averages of the monitored and simulated Y_F are 4.10 kWh/kW/day and 4.99 Wh/kW/day, respectively. All these values and others are reported in

Table 3. Monitored and simulated performance parameters.

	Monitored								Simulated
Month	YA	YF	YR	LC	LS	PR	CF	ƞSYS	YA	YF	YR	LC	LS	PR	CF	ƞSYS
Jan	3.04	2.95	4.30	1.18	0.09	68.69	12.30	10.03	3.96	3.89	4.39	0.43	0.07	88.7	16.21	12.94
Fab	3.72	4.05	5.19	1.47	0.05	78.05	16.89	11.40	4.56	4.48	5.14	0.59	0.08	87.2	18.66	12.74
Mar	4.5	4.41	6.56	2.02	0.13	67.25	18.38	9.78	5.31	5.23	6.21	0.9	0.08	84.1	21.79	12.29
Apr	4.78	4.79	7.12	2.32	0.15	67.28	19.95	9.83	5.94	5.85	7.14	1.2	0.09	81.9	24.37	11.97
May	4.78	4.62	7.27	2.49	0.15	63.59	19.26	9.29	5.81	5.71	7.16	1.35	0.1	79.8	23.79	11.66
Jun	4.85	4.84	7.50	2.65	0.16	64.56	20.17	9.43	6.05	5.96	7.67	1.62	0.09	77.6	24.83	11.34
Jul	4.77	4.60	7.46	2.69	0.16	61.68	19.18	9.01	5.75	5.66	7.54	1.79	0.09	75.0	23.58	10.96
Aug	4.23	4.21	6.56	2.33	0.15	64.19	17.54	9.07	5.61	5.52	7.24	1.63	0.09	76.3	23.00	11.14
Sep	4.47	4.46	6.74	2.27	0.15	66.21	18.61	9.67	5.15	5.07	6.37	1.23	0.08	79.5	21.12	11.61
Oct	4.22	4.09	6.03	1.79	0.14	67.87	17.03	9.91	4.76	4.68	5.67	0.92	0.08	82.5	19.5	12.04
Nov	3.39	3.38	4.77	1.38	0.11	70.92	14.1	10.36	4.28	4.22	4.89	0.61	0.06	86.2	17.58	12.59
Dec	2.90	2.81	3.98	1.09	0.09	70.41	11.69	10.28	3.76	3.69	4.18	0.43	0.07	88.3	15.37	12.89
Year	4.15	4.10	6.12	1.98	0.13	67.56	17.09	9.84	5.08	4.99	6.13	1.05	0.08	82.26	20.81	12.01
YA, YF, YF, LC and LS are with units (kWh/Kw/day)

and illustrated in Figure 14.

Some parameters are considered as key indicators, which can help to improve the management of power plants. In addition to the performance ratio and yields already detailed, it is essential to study the evaluation of losses (L_C and L_S), capacity factor (CF) and different efficiencies (η_SYS, η_PV and η_INV). Array losses (L_C) varied between 1.09 kWh/kW/day in December and 2.69 kWh/kW/day in July for the measured values, with an annual average of 1.97 kWh/kW/day. In comparison, the simulated values varied between 0.43 kWh/kW/day in December–January and 1.79 kWh/kW/day in July, with an annual average of 1.05 kWh/Kw/day. System losses (L_S) measured in December and January recorded the lowest value of 0.09 kWh/Kw/day against a maximum of 0.16 kWh/kW/day in June–July and an average of 0.13 kWh/kW/day. The simulated maximum L_S is 0.1 kWh/kW/day, and its simulated minimum is 0.07 kWh/kW/day, with an annual average of 0.08 kWh/kW/day. The variation in L_S compared to L_C increased from 3.6% in February to 8.72% in December and had an average of 6.59%; this explains the good efficiency of the inverter. The results are shown in Figure 15.

The measured annual average of the capacity factor (CF) during the year 2017 is estimated at 17.09%; it varied from 20.17% in June to 11.69% in December. In comparison, its simulated annual average is 20.82%, with a maximum of 24.83% recorded in June and a minimum of 15.37% recorded in December. Also, the measured efficiency of the photovoltaic module (η_PV) varied between 11.57% in February and 9.33% in July. On the other hand, the measured efficiency of the inverter (η_INV) varied between 98.54% in February and 96.27% in August. Therefore, the system efficiency (η_SYS) varied between 11.40% in February and 9.01% in July, and Figure 16 illustrates the three efficiencies.

6. Comparative Study

There are two major reasons to study a PV plant in detail. The first reason concerns the site studied, analyzing its performance and checking whether its operation is optimal and whether its economic yield is profitable. The second reason concerns the feasibility of future projects in similar climatic conditions and the mistakes to avoid; it is exactly in this context that this article falls.

Before making a comparative analysis of the performances of the LS-PVPP in Naàma (20 MW) with that of Ain Skhouna (20 MW) [23] and that of Adrar (30 MW) [30], it is important to draw up a climate profile for each of these stations.

The climate of Algeria is not uniform since the altitude in the country ranges from 19° N to 37° N. The northern part has a Mediterranean climate according to the Köppen classification and the rest of the country is considered a desert climate according to the same classification. However, there are microclimates specific to certain regions, as is the case of Saida (34°50′00″ N, 0°09′00″ E), which has a semi-arid climate. The altitude of more than 1100 m of Naàma, gives it a climate closer to that of Saida. Adrar (27.9° N, 0.31° E) is in a subtropical zone and is hot and dry; each year, it is among the hottest places on the planet in summer.

With the same annual average wind speed for Naàma and Saida, equal to 5.76 m/s, slightly higher than the average of 5.83 m/s recorded in Adrar, the three profiles illustrated in Figure 17 are very different. The wind curve in Adrar does not have specific extremes for a season or month. It varied from 5.37 m/s in September to 6.16 m/s in June. On the other hand, Saida had a windy spring, reaching 8.83 m/s, in contrast to a calmer winter of 3.49 m/s. Conversely, Naàma experienced a calm autumn of 3.8 m/s in October against summer wind speeds reaching 7.1 m/s in July and August.

Figure 18 shows the peak temperature recorded in July in the three stations: 31.97 °C in Naàma, 40.8 °C in Adrar and 33.11 °C in Saida. Winter in Naàma is the coldest, with 6.03 °C recorded in January, followed by Saida, with a minimum temperature of 9.13 °C recorded in February, and finally, a minimum of 14.06 °C recorded in Adrar in December. The geographical location and the decrease in rainfall imposed by climate change mean that Algeria experiences significant sunshine throughout the year, but especially in summer.

Adrar recorded its maximum irradiance during the period from May to July with an average of 247.65 kWh/m² and a minimum of 173.21 kWh/m² in January. Similarly, Naàma recorded a maximum of 227.28 kWh/m² during the same period, with its minimum irradiance recorded in December. On the other hand, Saida recorded a peak of 229.2 kWh/m² in June and a minimum irradiance of 113.17 kWh/m² in January. The three variations in irradiance are illustrated in Figure 19.

$${PR}_{Saida}PR=-0.519T_a+95.12(R^2=0.972)$$

(13)

$${PR}_{Adrar}=-0.73T_a+91.2(R^2=0.74)$$

(14)

$${PR}_{Naàma}=-0.36T_a+74.56(R^2=0.56)$$

(15)

Figure 20 shows a comparison between the average monthly performance ratios monitored for a year. The best PR is recorded in the Saida power plant where the maximum of 90.38% is recorded in January and February, while the minimum of 77.41% is recorded in July. The Adrar station recorded smaller values than the Saida station, with the same evolution but not the same gap. For example, the peak in January is 85.72%, with a gap of 4.66%, while the minimum is also recorded in July, which is 55.3%, and presents a gap of 22.11% compared to the minimum of Saida. As for Naàma power plant, it recorded a maximum in February with a monthly average of 78.05% and a minimum of 61.68% in July. Despite good irradiation in Naàma, the PR is reduced there because of the decrease in energy injected into the network, consequently affecting the system’s efficiency, as seen in Figure 21.

The highest value of the total losses of the three stations is recorded in Adrar, with a monthly average of 3.57 kWh/kW/day in July. Its minimum value of 1.07 kWh/kW/day is recorded in January and its annual average is 2.03 Kwh/kW/day. This average is lower than the annual average recorded in Naàma 2.1 kWh/Kw/day, which has, despite this, a maximum of 2.85 kWh/kW/day recorded in July and a minimum of 1.17 kWh/kW/day recorded in January. Saida recorded the lowest losses, which varied from 0.44 kWh/kW/day in January to 1.65 kWh/kW/day in July. In the case of Naàma’s LS-PVPP, the collection losses, Lc, are very high. The results are shown in Figure 22. The Lc values increase from March to October. This increase coincides with the increase in sandstorms that the Naàma region experiences every year due to drought. Dust has a double negative effect; the particles reflect light rays and prevent them from reaching the PV module, and humidity fixes dust on the module and prevents light rays from penetrating it, hindering energy generation.

The maximum recorded system efficiency η_SYS is 11.40% in February and a minimum of 9.01% is recorded in July, with an annual average of 9.84%. The simulated values of η_SYS are larger than those recorded over the whole year, with a maximum of 12.95% in January, a minimum in July of 10.96% and an annual average of 12.02%. The smallest deviation of 1.17% between the measured and simulated values was recorded for the month of February, which had the highest actual efficiency.

7. Photovoltaic Technology

Reference [31] presents a detailed study of a 9 MW grid-connected power plant in a semi-arid climate in India, similar to the Naàma site. The plant has three photovoltaic technologies, pc-Si, a-Si and CdTe, and monitoring is carried out on real data of 4 years of continuous monitoring from 2012 to 2015. A simulated projection is carried out over 25 consecutive years using PVSyst software in order to determine the impact of the semi-arid climate on the plant’s performance. The study calculated the following monthly average efficiencies: a CF of 18.4%, PR of 70.76%, PV eff of 10.1% and syst eff of 9.9%. A comparative performance analysis of the three installed technologies revealed that CdTe followed by the pc-Si technology performed better than a-Si.

Another study published in [31] addressed the effect of temperature in arid environments on PV modules of three technologies; monocrystalline silicon, polycrystalline silicon and thin films. According to the authors, losses in polycrystalline photovoltaic modules are greater than 50% when the temperature of the module exceeds 50 °C for hours. It would therefore be preferable to opt for thin-film modules, which are more adapted to heat.

8. Conclusions

This study focused on the 20 MWp LS-PVPP power plant, connected to the grid, installed in Naàma, a region with a semi-arid climate in the highlands of Algeria. Four meteorological parameters were considered, air temperature, irradiance, wind speed and humidity, based on data measured throughout 2017. The IEC 61724 standard, still in force, was used to frame this work, which was organized in stages. The first consisted of determining the parameters to be calculated, such as the energies produced and injected into the network, losses, yields and efficiencies. Then, a simulation was conducted using the PVsyst software, which made it possible to evaluate the gap between the measured and simulated values and the strong dependence of electricity production on climatic conditions. Afterwards, a comparative study was carried out with work published on similar installations in regions with a semi-arid climate or other conditions. The effect of climate change on the wind is already considerably visible in the Naàma region, as shown by increasing desertification and also by more contradictory extreme weather events such as floods and sandstorms. The solar panels are installed in an open area, known for its gusty winds during most of the year. Often, these strong winds turn into sandstorms. The 15° tilt of the solar panels favors the deposition of sand on the PV modules because of the humidity, which is quite high.

The simulated and measured wind speed values are very different over the 12 months, with a maximum difference of 60.56% for February and a minimum difference of 57.89% for October, knowing that wind strongly influences the ventilation of the solar panel. However, wind, which often carries sand, reduces irradiance and, consequently, the energy produced and energy injected into the network. As a result, the measured values for both parameters are lower than the simulated values, with a maximum deviation of 26.17% in August. The annual average deviation is 18.20%, which confirms that for the Naàma power plant, despite its semi-arid climate and sandstorms, its annual average measured value of the energy produced reached 26,195.96 MWh or 83.20% of its predicted value. The energy injected into the network was able to reach 25,371.86 MWh or 81.80% of its predicted value. During this study, it was noted that there is a lack of on-site information on the quality of the wind depending on whether or not it carries sand particles. The comparative performance study with the Adrar station showed that, for future projects, it would be preferable to use thin-film cells, which are less sensitive to temperature variations, to increase PV efficiency and reduce collection losses. The efficiency in the Naàma station is equal to 73.47% of the efficiency in the LS-PVPP of Saida. This is due to the location of this region, which does not experience sandstorms; moreover, it often rains there, which makes it possible for the PV modules to remain clean for as long as possible. While the temperature in Adrar has a negative effect on the station’s performance, it is an important parameter for the Naama and Saida stations. Naama’s performance is better in summer than Adrar, and the gaps in performance between summer and winter are smaller. Finally, it would be wise for simulation software to take into account the impact of climate change in recent years on the climate in semi-arid and desert regions in order to reduce the gap between measurements made in large-scale PV power plants and simulations carried out to validate the analysis of performance assessments.