Solar irrigation in sub-Saharan Africa: economic feasibility and development potential

To meet the crop evapotranspiration needs and close the irrigation gap in currently rainfed cropland (green water-irrigated only) of SSA, we estimate additional crop blue water demand of 67 km3 yr⁻¹, corresponding to a total yearly blue water withdrawal of 175 km³ yr⁻¹ (inclusive of losses) from surface and ground water bodies. Note that the blue water withdrawal (i.e., gross irrigation requirement) is significantly larger than the blue water demand (i.e. net irrigation requirement) because of water use inefficiencies in the withdrawal and in irrigation systems. This requirement is distributed over about 140 million ha (as derived from MapSPAM input data; figure 2) of smallholder rainfed harvested area (physical area equivalent for multiple growing seasons and crop rotation dynamics) of SSA where currently an irrigation water deficit occurs.

Zoom In Zoom Out Reset image size

Sub-regional monthly estimates in figure 2(A) highlight spikes in different months of the year that correspond to growing and simultaneously drier seasons of major crops: for instance, in East Africa the main sowing season of staple crops occurs around May and harvest takes place around October (light blue line's spike), while this pattern is anticipated by few months in highly cultivated areas of West Africa, such as Nigeria, Cameroon, and Togo (dark blue line's spike) [70]. To pump those water volumes and irrigate rainfed fields, we estimate 11 TWh yr⁻¹ of electricity requirements, corresponding roughly to the yearly power output of a 3 GW hydropower plant (assuming an average annual capacity factor of 40%).

The spatially explicit results of figure 2 allow identification of hotspots of crop evapotranspiration needs and solar pumping infrastructure requirements: among those areas, large areas in West Africa (28 km³ yr⁻¹, 5.7 TWh yr⁻¹, and 14 million solar pumps), mostly over Nigeria and over the Sahelian strip; the northern part of Mozambique and Tanzania (with 15 km³ yr⁻¹, 1.6 TWh yr⁻¹, and 9.3 million solar pumps required, respectively); the southern part of the DR Congo (7 km³ yr⁻¹, 0.5 TWh yr⁻¹, and 3.9 million solar pumps), and riparian areas of Lake Victoria (2.8 km³ yr⁻¹, 0.6 TWh yr⁻¹, and 1.6 million solar pumps), stand out. A summary of irrigation water and pumping energy requirements by country and by major crops is found in tables SI7 and SI8.

The sensitivity of results from relaxing the small field size constraint suggest (figure SI2(A)) that the estimated regional unmet crop water (and pumping energy) needs do not change drastically as a result of including large-scale cropland patches in the analysis, reflecting the dominance of smallholder farming in SSA. Specifically, field size data suggests that non-smallholder rainfed agricultural land only accounts for 15 million ha, or 10% of the total rainfed harvested area.

On the other hand, as seen from figure SI2(B), environmental flows preservation [71] (in terms of monthly withdrawals not exceeding local groundwater recharge and surface water discharge) represents a significant challenge for sustainably meeting the estimated unmet crop water demand, at least with the existing water infrastructure and management schemes assumed by the analysis. Specifically, we estimate that rainfed agricultural land where environmental flow sustainability constraints might be exceeded (see figure SI11) by less than 25% if irrigation gap closure was pursued without additional actions (e.g. cropping pattern change, land management solutions, fertilization) only accounts for 53 million ha, or 38% of the total rainfed harvested area. Thus, both infrastructure investment (such as reservoirs to convey and store water to mitigate seasonality dynamics) and water resources governance are deemed crucial complementary conditions for the sustainability of a widespread uptake of solar pumps (see section 4).

The resource and technology results presented so far describe the gross technical requirements under a baseline scenario, i.e. a status-quo analysis assuming current cropping patterns, climate, and water availability. In what follows, the economic feasibility of such physical requirements is assessed. At a region-wide scale, we estimate that over one third of smallholder farmers' unmet irrigation water needs in rainfed cropland (summing to about 60 km³ yr⁻¹ distributed over 55 million ha of harvested area, potentially satisfying 23 km³ yr⁻¹ of crop evapotranspiration needs) could be financially-sustainably equipped with solar irrigation, provided investment conditions are met. This represents about 40% of the total smallholder rainfed harvested area in SSA. This translates into a discounted investment requirement of $62 billion assuming a 20 year lifetime horizon (averaging at $3 billion yr⁻¹), in turn generating potential profits of up to $5.2 billion yr⁻¹, with the largest economic potential in Central and West Africa (figure 3(A)). Altogether, for SSA this corresponds to 11 million solar pumping systems with payback times of less than 20 years. The estimated feasibility areas, number of feasible systems, total investment needs and revenues and profit potentials are summarized table SI7 for each country included in the analysis.

Zoom In Zoom Out Reset image size

However, profit generation potential from solar irrigation adoption is unequally distributed across agricultural land (figure 3(B)), with about only 10 million ha of rainfed harvested area having potential to generate at least $100 ha⁻¹ yr⁻¹ of profits under current cropping patterns and historical crop prices. This figure grows to 20 million ha of land if a threshold of $50 ha⁻¹ yr⁻¹ is considered. 30 additional million ha show very little profit potential by solar irrigation adoption only, whilst in the additional 100 million ha rainfed harvested area of SSA (not plotted in figure 3(A)), solar irrigation is not found to be economically feasible, at least without further action (e.g. fertilization and cultivated crop shift).

The necessity of lowering risks and the cost of capital is confirmed by a sensitivity analysis on the impact of considering different discount rates (7.5%, 25%, and 40%) than the baseline (15%)—presented in figure SI3. The results show that the discount rate has a significant impact on the lifetime costs, revenues, and profits of potential solar pumps in SSA, and therefore on the number of sites where solar irrigation is estimated to be economically feasible, ranging from around 11 million pumps and $5.2 billion yr⁻¹ of profits under a 7.5% discount rate down to 8 million pumps and about $2.5 billion yr⁻¹ of profits under a 40% discount rate.

Sensitivity analysis for the change in the cost of PV (break even cost with diesel, thus suggesting e.g., a change in the price of diesel, as well as cost change in PV manufacturing costs) reveal the great importance of such cost components (figure SI6), and thus also the potential of incentives. For instance, a 10% reduction in PV & battery costs leads to a near doubling of the number of economically feasible pumps (from 11 to almost 20 million), irrespective of an only marginal decrease in yearly discounted costs. Moreover, sensitivity analysis for crop price variability (an important variable given commodity prices volatility) reveals (figure SI7) that potential profits (and thus economic feasibility) from solar irrigation are also rather sensitive to crop prices. Compared to the baseline scenario of 10 year median prices, 10 year minimum prices imply 35% lower profits and nearly 2.5 million less economically feasible pumps. Conversely, of 10 year maximum prices benefit the feasibility assessment, with 1.3 additional solar pumping systems and 25% higher profits.

As illustrated in the methods, we operate additional scenarios to assess the relevance of business models for the economic feasibility of solar irrigation. The results—presented in figure SI4—reveal that incentives have a notable impact on the number of economically feasible sites. Overall, a business model amortising all upfront costs more than doubles the number of feasible solar irrigation systems, with incentives on the PV system representing the key drivers of such observed impact. This is a crucial finding—consistent with previous literature contributions [72, 73]—highlighting the need of lowering upfront barriers if decentralized solutions are to become widespread in SSA. Finally, we also evaluate an array of other sensitivity scenarios, including consideration and exclusion of battery storage from the PV system, solar PV value added tax (VAT) and import costs exemption, as well inclusion of a water storage tank (see SI).

When looking at the spatial distribution of economic estimates (figure 4), we find solar irrigation to be feasible and profitable in large part of the southern Democratic Republic of the Congo, the Congo Republic, vast areas of Nigeria, regions along Sahel, as well as croplands in Tanzania and Malawi. Other scattered feasibility areas are distributed across the continent, e.g., districts of Kenya, Ethiopia, Zimbabwe, Madagascar, Angola, South Africa, South Sudan. Conversely, sites found not to be suitable for solar irrigation (figure 4(D)) consist of areas where either water sources are hard to access (e.g. deep groundwater wells and remote surface water sources), PV potential is reduced, currently cultivated crops would not benefit substantially from the input of irrigation systems in terms of yield response and thus revenue generation potential, or remote areas where overall costs are higher than potential revenues.

Zoom In Zoom Out Reset image size

Finally, it is relevant to examine the distribution of the modelled technological and economic indicators to understand the range of values and variability that emerge in different locations where solar pumping is found to be an economically feasible investment.

An analysis of the distribution of key indicators from the local techno-economic analysis (figure 5) suggests that the mean size of a pumping system in our analysis is about 0.6 kW, with a solar module of 1.8 kW and a battery of capacity 3.5 kWh. This translates into a mean lifetime discounted system cost of about $7600 (of which $4200 for the PV system; $1850 for the pump; $1150 for the irrigation system and farming costs; and $430 for transport) in turn generating mean revenues and profits for around $4100 and $930 yr⁻¹, respectively. Notably, we calculate a mean utilization rate of about 75 days per year (under the modelling assumption that irrigation is performed every second day during the cropping season). This usage pattern translates into a mean electricity consumption of about 255 kWh/pump/yr. Finally, the mean system payback time of economically feasible sites is estimated to be below 10 years.

Zoom In Zoom Out Reset image size

To complement the techno-economic analysis, we estimate potential co-benefits of large-scale solar pumps adoption, restricting this assessment to areas where solar irrigation is estimated to be economically feasible. Firstly, to determine co-benefits to SDG2, we calculate that the transition could positively impact food security, generating an additional 187 kcal and about 3 and 7 protein and fat grams per capita per day, respectively across SSA (based on FAO representative crop nutritional contents, table SI5). Considering that these are regional average values, these represent significant gains if compared to the average requirements of 2000 kcal/day, 50 g proteins/day, and 60 g fats/day [74]. Such potential gains are also very relevant from a food self-sufficiency point of view, which is a strategic priority for many developing countries.

As seen from figure 6 (the numbers are also summarized in table SI9), we find substantial inequality in terms of food production growth thanks to solar pumps adoption across SSA countries, with some nations, such as the Republic of Congo [COG] (+2000 kcal/person/day), Tanzania [TZA] (+700 kcal/person/day) and Guinea [GIN] (+440 kcal/person/day), showing significant potential to close their national caloric gaps and even increase their food exports, while others with substantial food gaps, like Somalia, Zimbabwe, Liberia, Central Africa, Republic and Uganda (see table SI9), requiring larger imports than the estimated yield growth potential to achieve national food security.

Zoom In Zoom Out Reset image size

Finally, across SSA, we estimate more than 33 GWh/day of potential residual power (electricity output from economically feasible solar pumping systems' PV module which is not used for water pumping) i.e., about 12.5 TWh yr⁻¹, distributed as shown in the maps in figure SI10. To give a reference, the yearly total (i.e. inclusive of all sectors) final electricity consumption of Nigeria, the most populous country of SSA (206 million people, nearly half of which live without access to electricity), is 27 TWh yr⁻¹ [75].

Figure 7 (panel (A)) demonstrates that the total unmet crop evapotranspiration needs in currently rainfed cropland range from about 67 to about 95 and 100 cubic kilometres per year in SSP245 and SSP585 scenarios in 2050, respectively. In addition, with growing climate change impacts the proportion of sites where solar irrigation is not found to be feasible for economic or environmental barriers grows (figure 7(B)) from 64% to about 70%. An unbounded pumping scenario displays significantly larger (82%) share of feasible sites and groundwater resources exploitation than an environmental flows preservation scenario (55% of sites).

Zoom In Zoom Out Reset image size

In terms of the implied electricity demand for water pumping in the three climate scenarios for the two variants (figure 7(C)), we observe a striking difference in pumping rates (and thus energy consumption) between an unbounded pumping and environmental flows preservation scenario (from 11 to 0.5 TWh yr⁻¹) because of potential overexploitation of ground water aquifers and surface water sources. Moreover, we estimate a considerable energy demand growth with climate change compared to under historical climate conditions observed in the unbounded pumping scenario (growing to 14–15 TWh yr⁻¹).

To conclude, we quantify the monthly gap between the groundwater withdrawal needs for irrigation gap closure and the maximum amount of extractable water, showing that the gap grows substantially with growing global warming (from 57 to 61–76 km³ yr⁻¹), and reaches critical levels in high demand periods (figure 7(D)), e.g. from 9 to 14–17 km³ in August. Finally, the analysis reveals that both the harvested area extent suitable for solar irrigation and the additional potential food yield decline (from 55 to about 45 million ha and from 73 to 59–61 trillion Kcal yr⁻¹, respectively) under warming climate futures, reduces food security and development prospects (figures 7(E) and (F)).

Our study estimates that SSA is facing an unmet blue water demand of 67 km³ yr⁻¹ over smallholder farmed rainfed cropland for the 19 crop types considered. This estimate compares well with previous studies, e.g. the value obtained by Rosa et al [30] under a baseline scenario assuming irrigation expansion over rainfed areas to cope with water stress and increase production and, thus, the number of people fed. Our analysis suggests that over one third of such unmet water needs—distributed over about 55 million ha of (harvested) area—could be supplied with solar irrigation (with 10 million ha with a profit potential >$100 ha⁻¹yr⁻¹). This translates into a requirement of 11 million solar pumping systems. For reference, Indian farmers currently irrigate their fields with more than 30 million agro pump-sets [76], of which about 8 million are off-grid. Of those, according to the most recently available survey, about 250 thousands are solar pumps, and the Indian government has set an ambitious objective of achieving 2 million solar pump installations by the end of 2022 [77].

To finance those installations, a region-wide cumulative discounted investment requirement of $62 billion is estimated assuming a 20 year lifetime horizon, in turn generating potential additional (on top of baseline crop yield) revenues of over $8 billion yr⁻¹. Areas where solar pumps cannot be paid back within 20 years are not deemed suitable and thus excluded from these gross figures. In addition, we estimate that the transition could positively impact food security and general access to energy services. Altogether, these results suggest that solar pumps bear significant economic feasibility potential. This goes in the same direction of previous analysis, e.g. Dalberg and Efficiency for Access Coalition [78] estimates that the market for solar water pumps in SSA will expand to as many as 2.8 million households and a value of $1.6 billion per year by 2030.

While electricity access is crucial directly on farms, it is also a core issue (SDG7.1) for most other sectors in rural SSA, and primarily for residences, education, healthcare, and small and medium enterprises. In our analysis, we size PV systems based on the site-specific water pumping needs. However, on certain hours of the day and seasons of the year when irrigation is not required, such PV systems might be employed for other purposes, especially if PV modules are transportable. We estimate about 12.5 TWh yr⁻¹ of residual power output not used for pumping and potentially usable for other energy services such as crop processing and home uses, provided appliances are available to households and farmers. Nonetheless, it is important to bear in mind that this excess PV output is unevenly distributed over space and time depending on the irrigation schedule and is thus likely not sufficient to cover all needs at home and on the farm for farmers. Nonetheless, it might provide an important first step along the energy ladder [79, 80] and enable some additional energy services such as raw crop processing.

Once techno-economic potential is demonstrated, the challenge moves to the implementation side. In fact, despite promising prospects [39, 81–84], the uptake of solar pumps in SSA is still very low. The current implementation of solar irrigation in many parts of SSA is driven by donors including European countries, non-governmental organizations, World Bank, and other UN agencies.

To put solar irrigation infrastructure on the ground at a large-scale and achieve the potential estimated in this study, private capital (also as part of public–private partnerships) is indispensable. As demonstrated by recent research [72, 73] and through our business models simulation analysis, upfront costs, capital cost, and private discount rates represent a key barrier for successful large-scale uptake of decentralized energy infrastructure in the region, including solar irrigation systems. In turn, these factors depend on the quality of national and regional regulatory frameworks and institutional arrangements. To achieve rapid solar technologies uptake in SSA and mirror examples such as India [85], techno-economic potential is in fact not sufficient. Public–private local research and development (R&D) programs [85] in the sector—both nationally and regionally—are a necessary condition for tapping the estimated techno-economic potential: enabling regulatory, market and governance conditions are in fact crucial to ensure a lower cost of capital and market penetration of private capital for decentralized service supply technologies [73].

It is the responsibility of public decision-makers to create the right policies, incentives, and investment environment for private companies to develop, install, and manage this infrastructure and deploy solar irrigation on a large scale and farmers to invest and gain capacity to use these. Future uptake will largely depend on government subsidies and regulatory reform [86, 87], as well as on the use of smart business models [88] by solar pump supplying companies, as it is estimated that it generally takes 6–12 months of income for a typical farming household to cover upfront system costs [78].

Another key challenge to ensure increased solar pumps uptake and sustained utilization is the consideration of technical knowledge and materials availability, including technical skills to be able to repair the system when it breaks, as well as social norms and structures, such as the acceptability of solar irrigation and the establishment of shared ownership models for groups of farmers and smallholder consortia. This goes beyond the sole infrastructure uptake issue, but it also includes consideration of capacity development over farming practices, including irrigation management, fertilization, pest control, crop rotation, and extensification–intensification trade-offs. Only when all these dimensions are accounted for by both public decision makers promoting a transformation of the agricultural system and private retailers providing and installing systems can a successful uptake and positive development impacts be experienced.

Finally, besides uptake and private use of solar pumps by farmers, several institutional and socio-environmental aspects are of great importance, despite being beyond the scope of this paper. Irrigation infrastructure, in particular when sprinklers and pumps are used, requires intensive maintenance and water withdrawals need to be monitored by dedicated public authorities to avoid 'tragedy-of-the-commons' issues [89] such as overuse of water, declining groundwater tables, and salinization. Sustainable irrigation requires strong institutions responsible for developing and enforcing rules to avoid unsustainable water use practices, which will be a key development priority for SSA policymakers [90].