Sunday, February 14, 2016

Grid-Connected Solar Energy Projects in Botswana

Dumelang*. In my previous post, I took a good hard look at why there is so little photovoltaic (PV) solar power generation in Botswana and shared with you some of the complexities involved in implementing large-scale solar here. Botswana does have an impressive solar resource, but its exploitation requires a great number of tradeoffs. As a result, there are a limited number of larger-scale solar systems in the country. In this post, I focus on the grid-connected operations. These are systems that feed power back into the grid or that use the electrical grid as a backup, i.e., the grid makes up for any shortfalls in energy generation by the PV unit. I will look at off-grid systems in my next post.

There are presently three large grid-connected systems in Botswana: a single large-scale 1300 kW solar farm in Phakalane to the north of Gaborone; a recently constructed, but not yet operational, 20 kW EU-funded University of Botswana research system installed in Mokolodi village, just south of Gaborone; and a 34 kW system, owned by Scales Associates and located in the Broadhurst area of Gaborone near the Western Bypass. There are a number of other small-scale installations with similar configurations, i.e. grid connected but just using the grid as backup,  that have been installed on residences in Gaborone and surroundings. This post takes a close look at the first two of these grid-connected projects.
   
The Phakalane 1300 kW Solar Power Plant

On the southern outskirts of Phakalane, some 15 km north of the capital, near an industrial area and partially surrounded by undeveloped land, sits the 1300 kW project (see Google Earth view below). The facility is located on 2 hectares and is surrounded by a sturdy fence supplemented with electric fencing and rolls of razor wire and there is a guard permanently on duty at the gate. An informative signboard provides information about the project and a diagram shows the main components.


Phakalane Solar Farm Information Board

The project was completed in 2012 and electricity generation started in August that same year. The projected was largely funded by P80 million  from the Japanese government and some contributions from Botswana Power Corporation (BPC) and the Government of Botswana. The total cost of the project was approximately P106 million (approx. US$13.4 million). Even at the time that the facility was built, these project costs were on the high side. Using an average 2012 exchange rate of P7.90/US$, the cost comes to $10.30/watt (W). Average costs for solar projects in other parts of the world were of the order of $3/W at this time. (In 2013 and 2014, utility-scale PV projects in neighboring South Africa were being installed for ~$3/W.)  Now, normally the first major project in any country is an expensive one because everything is new and more ground work has to be undertaken, but a price greater than three times international averages seems exceptional. This most likely contributes to the prevailing perception in Botswana that solar PV is expensive.

The system contains 5920 panels, each with a 220-W DC rating, which gives 1 300 000 W or 1300 kW overall rating. The panels are wired in strings of 16 panels connected in series to provide a peak voltage of 470 V DC. These strings are then combined in parallel with other similar strings and fed into combiner boxes and then into inverters. The electrical output of solar panels is always direct current (DC), but what we transmit over power lines and use in our homes is alternating current (AC), so the function of the inverters is to convert DC to AC. These inverters output 380 V AC, which is then ramped up to 11 000 V through a transformer so that it can be fed directly into the BPC power grid.

I am always impressed that an array of solar panels, each with an output of ~30 V DC can be combined, inverted, and transformed into a grid-compatible AC voltage of 11 000 V. For electric power engineers this is old hat, but I am always impressed. I have visited numerous power plants and am used to seeing large machines thundering away, devouring large quantities of fuel, and spitting out large volumes of waste heat, pollutants, and carbon dioxide to produce electricity. It is a therefore a pleasure to see a field of solar panels, each quietly converting sunlight to electricity and then using technology to configure wiring, inverters, and transformers so that the electricity can be boosted to a form that can be transmitted over the grid and used to power homes in other parts of the country.

In the energy field, one needs to be sure to understand what is meant by the rating of a power plant. Most power plants, say the 600 MW coal-fired operation at Moropule B, refer to their output of AC electricity. In this case, it is easy to calculate how much energy a power plant would generate over a certain time period. For example, if all the generators at Moropule B were running at their rated output, uninterrupted for 24 hours, the output would be:

600 MW x 24 hours = 14 400 MWh = 14 400 000 kWh (1000 kWh = 1MWh)

However, with solar PV operations, the rating is most often given as the combined DC output capacity of the panels under the standard irradiation condition of 1000 W/m2 at 25oC – the conditions known as one peak sun (see an earlier post for an explanation of irradiation and the peak sun hour concept). Although the amount of sunlight in Botswana is high relative to other parts of the world, the irradiation levels are only close to one peak sun at around noontime. A solar panel will therefore only produce its rated output for a short while around midday; the rest of the time, the irradiation is lower and the output is commensurately lower. The figure below shows recent irradiation levels for Gaborone, as measured by a weather station on the roof of the Faculty of Engineering and Technology building at the University of Botswana.


Because irradiation conditions change during the day, we determine the energy that can be potentially harvested in a 24-hour period by calculating the area under the irradiance curve—the yellow area in the figure above. This measure of energy is termed insolation; it is measured in kilowatt hours per square meter (kWh/m2). Another measure of insolation is to calculate how many hours of peak sun (with a fixed irradiance of 1000 W/m2) will deliver the same energy as the sum of the varying (irradiation x time) values over a day. The number of hours of peak sun per day is a particularly useful measure and is extensively used in the solar energy field to determine the daily output of electricity from solar panels. A typical insolation value for the Gaborone area is 6.09 peak sun hours. 

Cloud cover significantly affects the performance of solar panels: the figure below shows the contrast between a day with no cloud cover, February 1,  a day with lots of intermittent cloud cover, February 7 and Jan 14, which was overcast for the whole day: even with just passing clouds, the drop off in irradiation levels is significant.




In addition to a limited number of hours of peak sun per day and cloud cover, there are losses through the electrical system: in the wiring; in the conversion of DC to AC in the inverters; and through the transformers. Losses in PV systems are typically of the order of 20 to 30%.

There are also performance losses due to temperature. Intuitively, one would expect hot sunny days to be ideal for solar power generation, but one aspect of PV technology not often appreciated is that electricity output of PV panels actually decreases as the temperature increases by approximately 0.5%/ oC. Solar panels are rated for a temperature of 25oC. Panels operating in the Gaborone area, where the temperature on a summer’s day can be 35oC, can expect a 5% decrease in performance. Furthermore, when the clouds roll over the skies during the rainy season, we can also expect a big decrease in electricity production.


The European Commission’s Institute for Energy and Transport (IET)  has developed a great online calculation tool for estimating average losses and outputs for PV systems. One enters the geographic location, the rated power of the installation, and panel orientation and some very useful information on average irradiation, system losses, and PV output are generated. Alongside is the information I generated for the Phakalane location using this calculator.

As can be seen, the average losses were calculated to be 27%, the average daily peak hours were 6.49, and the average output was determined to be 6150 kWh/day, which totals 2240 MWh AC per year. With average systems losses of 27%, the output AC rating of the plant would be 819 kW.

During a recent (January 2016) tour of the plant, I was able to gather some information about its operation. It started up in August 2012 and since then has operated with only four months of downtime (June 2014, July 2014, Dec 2015, and Jan 2016), so the plant has operated for 36 months or 3 years. During this time, the plant has generated a total of 5286 MWh of electricity.

According to the IET calculator, the total electricity output over this time period should have been 3 years x 2240 MWh/yr = 6720 MWh, which means the plant has generated 79% of its expected output. The differences are due to: 1) lower solar irradiation levels during this period – data from the monitoring equipment at the Phakalane plant indicated average monthly irradiation levels of 161 kWh/m2 – rather than the 197 kWh/m2 used in the calculator; 2) the effects of cement dust from a nearby brick plant coating the panels and reducing their efficiency; and 3) a higher-than-estimated temperature effect on the panels. It always astounds me how hot solar panels get when they are exposed to sunlight: surface temperatures of solar modules can rise to greater than 70oC – much higher than ambient temperatures. At 70oC and with −0.5 %/oC performance losses, decreases in performance can reach as much as 23% at certain times during the day.

Another useful calculation is to determine the capacity factor, which is a measure of the proportion of its maximum output that a power plant actually delivers over a year. A capacity factor of 100% assumes that a power plant runs for 24 hours a day, 365 days a year. Coal-fired power plants have capacity factors of about 60%, nuclear plants 90%, wind plants about 35%, and solar PV plants are usually of the order of 26%. If we assume that the Phakalane plant has an output of 819 kW (assuming 27% system losses), we can calculate that over three years of operation:

Capacity Factor = 5 285 860 kWh / (3 yr x 365 d/yr x 24 h/d x 819 kW) = 24.6 %

This is in line with that expected for solar PV operations.



The Phakalane project is a well-designed operation and, like other PV plants, requires relatively little in the way of maintenance and operating costs. Costs include some electrical component maintenance, security services, cutting back of vegetation and weeds so that panels don’t get shaded, and the occasional washing of the panels. Unfortunately, it is located near a cement and brick factory and, as a result, there is a carryover of fine cement dust that settles on and coats the panels, especially those closer to the factory: a fair amount of effort is required to remove the encrusted coating on these panels.


The single biggest challenge for the Phakalane plant is vandalism and panel theft. There was recently intrusion at the site and large amounts of grounding cables were stolen for their copper value. Six panels were also stolen, despite the installation of special anti-theft fasteners on the panel attachments. As noted previously, the facility has been down since December 2015 and during my tour in mid-January I noted that repairs were underway. Hopefully the plant will be operational again before long.


As solar installations proliferate, panel theft is becoming a growing problem worldwide. Many installations are located in remote areas with limited security, so it is often easy for thieves to access the area and take their time to unbolt and load up the panels. Added to this, solar panel installations have large amounts of grounding wires, making them an inviting target for thieves who are looking to steal copper wire that can be sold to unscrupulous scrap dealers.

University of Botswana’s Mokolodi Village Project

To the south of Gaborone, and bordering the Mokolodi Game Reserve, is the small village of Mokolodi, which is home to about 600 people. This is a relatively new village by Botswanan standards and is surrounded by large properties and some high-end housing developments. Right in the heart of this village, the Clean Energy Research Centre at the University of Botswana has built, using money from an EU grant channeled through the African Union, a small, but in the history of solar power in Botswana—an important solar PV installation.

This will be the first of its type in Botswana to feed electricity produced in excess of that used onsite into the local low-voltage grid at 400 V. Phakalane is also a grid-scale project, but it is designed to feed all of its produced electricity into the high-voltage section of the national grid, whereas Mokolodi is designed first for local energy use and then for directing excess energy into the grid. It is important to appreciate that feeding excess solar power onto a grid is not a cutting-edge concept: in fact, it is the basis for most smaller solar projects installed in the US and Europe during the past decade or so. These systems are designed to generate electricity during the day to power the home or business; any excess generated is fed back into the grid and electricity is drawn back from the grid during the night hours as needed. Most of these systems are not designed with battery storage, so they are not self-contained off-grid systems: the grid is their backup system and they are very dependent on the grid to compensate for electricity needs at nighttime or for shortfalls during cloudy days.Another important aspect of these types of systems is that they are not a buffer against the effects of load shedding or other grid interruptions: when the grid goes down, they are designed to switch off immediately. This is to avoid little islands of solar electricity generation feeding electricity into a dormant grid and that could pose an electrical safety hazard for any electrical line workers who might be repairing lines in the area.

The Mokolodi system has a rating of 20 kW and, because it is an experimental system, consists of the following components:
  • A 5 kW system on the village clinic to provide daytime power for the clinic and to feed excess electricity into the grid;
  •  A 2 kW system on the home of the village Chief to provide daytime power and to feed excess electricity into the grid during the day;
  •  A 10 kW system at the Kgotla (meeting place), designed to power the Kgotla and some Village Development Committee owned homes during the day, with any excess fed into the grid;
  • A 3 kW experimental system that consists of three different types of silicon-based panels that their performance can be compared and contrasted;
  •  A small weather station to measure the solar irradiation, temperatures, and wind speed.

One of the more interesting aspects of this project is the direct comparison of the performance of the three types of panels (1 kW of monocrystalline silicon panels, 1 kW of polycrystalline panels, and 1 kW of amorphous silicon panels) under the energetic and harsh Botswana sunlight. UB researchers will be able to monitor their performances over the short and long term to determine which would be best for the high irradiation conditions here. The photographs below show some views of the Mokolodi project. 

13 kW University of Botswana Solar Installation at Mokolodi Kgotla


Once the project is up and running, the UB team is hopeful that the project will provide the following benefits:
  •  There are presently several non-electrified homes not connected to the grid. These homes will receive “free” solar power from the grid during the day; at night, they will draw power from the grid and pay normal BPC rates.
  • This is a demonstration installation for similar low-voltage grid-connected projects. It will allow BPC and the UB team to gather information and experience and open the way for the installation of similar systems throughout Botswana.
  • The generation of any extra power will be feed into the grid at no cost to BPC and, in a very small way, reduce the amount of coal that needs to be burnt.
  • It will be a valuable research tool for UB and its students and will allow the in-depth investigation of small-scale solar systems and their variability, their impact on the grid, and the performance of different panel types.
The UB team at the Clean Energy Research Centre will also be investigating the socioeconomic benefits to communities of free solar energy during the day and purchased electricity at night. The project is close to completion and hopefully will be commissioned in the near future. As the project moves forward and starts generating research results, I hope to share some details in a future blog.

This post has taken an in depth look at two important grid-connected PV systems in Botswana and in the next, I will be discussing the far more numerous off-grid systems that are scattered throughout the country. In the meantime, look out for solar projects and remember to turn off the lights when you leave the room.

Tsamayang Sentle**
Mike Mooiman
mooimanm@franklinpierce.edu


(*Greetings in Setswana)

(**Go well or Goodbye in Setswana)