This is a theoretical business plan. It attempts to specify the characteristics we need and the things we don’t know, as well as a lot of assumptions on which we are basing the resulting calculations of cost. The goal is to lay out a strategy for penetrating the rural Africa market for electricity, thought by many to be the next big opportunity for private investment in the power sector. It’s not clear that this “many” have ever actually tried to do a project in Africa, but never mind that.
- One of the seven major development goals adopted by the United Nations in 2015 as “Agenda 2030” has to do with energy. It states as the 7th of 17 Sustainable Development Goals:” Ensuring access to affordable, reliable, sustainable and modern energy for all by 2030.” Most of these “all” who don’t have this access live in Africa.
- Africa (by this we mean the 39 countries south of the Sahara, “SSA”) is falling behind, not getting ahead, in providing basic electricity to all of its people. According the IEA statistics, in 2009 there were 585 million people in SSA without electricity, and in 2014 that number had risen to 632 million.
- Two thirds of SSA citizens don’t have access to electricity, and these people are, no surprise, not in cities. They’re not rich either, so the price of electricity to be delivered to them must be economical. But reliable. Repeat “economical.”
- There is widespread agreement that access to electricity in rural areas has a number of positive developmental benefits: it makes farming more efficient (think water pumping for starters, and refrigeration), raises education levels, lowers health risks (no more charcoal cooking indoors), and lets people start small businesses that are otherwise not possible if done by hand and candlelight—clothing manufacture, for example.
- Despite the need, big projects have not worked well, whether proposed by government owned utilities, IOU’s or Independent Power Producers. The causes are many: corruption, bureaucracy, ineffective governments and government agencies, under developed markets, bad credit and lack of capital at all levels, poor quality local workforce, use of high cost imported machinery, and on and on. One can also include war, massive civil unrest and failed nationhood.
- Sheer size in power projects is an obstacle despite some economies of scale. Power Africa, a USTDA initiative, has in four years of effort only managed to facilitate the connection of two million customers, according to its Annual Report. The best demonstration of this problem is, or perhaps was, a proposed dam on the Congo River in the Democratic Republic of the Congo. The third of the three proposed dams, Inga III, would have produced 39,000 MW of power and cost an estimated $80 billion. Never mind that the largest dam in the world, Three Gorges in China, is 22,000 MW. Its development and construction was heroic, even for the Chinese. And many argue that its economics have never really proven out. And never mind that the last dam built in the Congo was completed in 1982, 35 years ago, after 14 years of construction. Selling power from a large dam in the DRC, given its status as a non-functioning state, would be tricky. No, that’s not right, it would be impossible. I am not even sure you could give away electricity there, since the distribution system is close to completely broken.
- Rooftop solar approaches are not effective either. Costs are too high, financing mechanisms don’t exist, individual owners have neither capital nor credit, and even for solar there are individual design issues and interconnection issues. Moreover, rooftop solar, because of its small scale, custom nature and inability to take advantage of such economies of scale as there are, is approximately three times the cost of utility scale solar. But “utility scale” can be as small as three or four megawatts and still be cost effective.
There have been several articles recently trumpeting a decentralized or minigrid approach to rural electrification. Unfortunately these are generally sparse on actual data. An article in the September/October issue of North American Clean Energy magazine happily announces in its title: “Microgrids—Bringing power to remote regions of the world.” It then describes several projects, but mentions nothing about costs. It cites two projects, one of 93Kw in Haiti and one of 191 Kw in western Colombia that “provided reliable electrical power” to 450 and 431 households, a peak capacity per customer of 206 watts (2 light bulbs) and 443 watts (4 light bulbs) respectively. The customers were not asked if they thought this was reliable electric power, esp as the power was exclusively from solar panels, with no batteries. The article states that the microgrids provide electricity for “most of the day.” It also then declares, “Children can now study at night, parents can cook in the evening…and businesses can stay open after the sun goes down.” How this is achieved is not detailed.
An article written by Bill McKibben in the 26 June 2017 New Yorker, entitled “The Race to Solar-Power Africa” is a bit better, but only a bit. The subtitle says, “In eighteen months, entrepreneurs brought electricity to hundreds of thousands of people in places that the grid failed to reach.” The article tells stories about two different companies, and cites current costs for kerosene and cell charging as ranging from $19 to $30 per month per household, but does not indicate where the “hundreds of thousands” are. Both companies are in essence rooftop companies.
A somewhat more realistic article in the IEEE Spectrum blog, titled “Off Grid Electrification Financing is Failing” comes to the conclusion telegraphed in the title. Politics, vague planning, and lenders unwilling to support small projects share the blame.
We believe that there is a “middle way” that can be tried in African rural communities to promote more rapid electrification than is currently occurring. We suggest taking a page out of the history books, in this case the US history of a depression era program called the Rural Electrification Administration.
The REA was established by President Roosevelt in 1935 to address the fact that 75% of rural Americans had no access to electricity. And given the era, this was a significantly larger portion of the whole US population—almost half. The REA did two things to solve the access problem: it focused on developing rural nonprofit cooperatives, owned by the people whose power the co-op provided, and it funded these entries with low cost government credit. It did not serve as an extension of the existing gird and its for profit owners, but it instead created rural, isolated islands of power generation and distribution. Eventually these small grids were connected to the larger grid, but not initially or during their development. As an example, a coop in the middle of Oklahoma, REC, started with 100 customers. The result of this and other rural coops was that the connection numbers went from 25% in 1933 to virtually 100% in the early 1960’s. The REA’s electricity job was done by then, although the agency, having created a strong rural constituency, has hung on, continuing to fund the needs of co-ops, and adding telephone service to its mandate.
Some have suggested that this is a fine model for rural Arica, emphasizing the provision of credit, a real need in the area. However, there’s no FDR equivalent leader in Africa, and African countries have great difficulty in even funding the needs of their existing utilities to create reliable systems. The genius of the REA was not only cheap money, it was the recognition that small electric companies could be established effectively and economically. And thanks to significant changes in the cost of both solar and battery technologies, that opportunity exists in Africa today.
There is one very important difference from what many companies are trying today with minigrids. We believe that you cannot effectively cover the development costs, the so-called “soft costs,” of very small projects. These costs—finding and negotiating land, permitting, contracting for construction, etc.—are more or less invariant with the size of the project. Spreading these costs over more Kw or Kwh is key to holding total costs down. You need some reasonable scale to make the economics work. We believe that this number is between 2 MW and 10 MW, but it would be dependent on the community or communities to be served. This size project could deal with planning/acquisition/permitting and system design costs because it would be able to amortize these real and inescapable costs over a larger customer base.
Consider for a moment this comparison: would a developer rather develop a 500KW minigrid ten times, or a 5 MW minigrid once? Just think about land acquisition, negotiation of sale or land lease, verification of title, registration of sale, etc. And government approvals. And financing. The transaction costs of repeating each step ten times would consume the small project’s economic value.
We advocate designing and installing an entire stand-alone “rural electric co-op” in each of a number of appropriate rural communities, not as any part of the centralized gird. One obvious implementation suggestion: A thousand customer system will not materialize out of thin air, and will require significant capital deployed upfront so that the mini-utility is ready to hook up customers and deliver power before it starts any sales activity. This is one disadvantage compared to the roof top by roof top approach. It would be best if one could start with one large customer as a sort of “anchor tenant” and then add other smaller and residential customers as demand increased. Perhaps a hospital, a clinic, a school, a small manufacturing plant, a market, or some other established entity. Each situation will be different, but the principles should be the same. And the results will justify further expansion of the model.
Each co-op would be composed of three components:
- A 5 MW solar array and a 5 Mw/5Mwh battery installation, each designed in a modular fashion. All the land and permits for the full system would be acquired and the system designed, but the individual modules might not all be installed at the beginning. This would allow for the size of the batteries or the number of panels to start small and grow over time as necessary. Little will be known at the beginning of operation; how much electricity will people use if they have never used any before? Sizing needs to be flexible. As demand increases, batteries can be added if the control hardware is configured appropriately. Note that this is a larger ratio of battery to solar capacity because here the battery must not only deal with the instantaneous cloud based intermittency of the solar panels, but also must provide the carryover evening power. Hence it needs to be larger than more traditional US and European “solar + battery” projects. It is also likely that a small diesel generator will need to be provided as backup for those periods when the battery has not been able to be fully charged by the panels during the day, due to weather. But the generator will only be run at its optimum set rate, and to charge the battery, not to provide direct power to the system.
- A delivery system, assumed to be at distribution voltage (110 or 220, depending on the dominant voltage in the country) with wires, poles, transformers and safety/protective equipment, but no high voltage links—and thus much cheaper than bigger systems incorporating high voltage segments. We don’t need high voltage transmission lines since this is a small system and needs to be simple and inexpensive. Distances should not be long but several villages may need to be aggregated. We also do not plan to provide metering to customers, but instead charge them a flat rate. So no meters and no meter readers. As usage increases over time, it may be useful to install meters, but not at the beginning. We do not envisage connecting this to the “central” grid for some time, but instead running this co-op on a self-sustaining basis. The key difference in this system is it recognizes that volumetric pricing (cents per Kwh) doesn’t make sense when the Kwh are in essence free. If the project is getting its capital back with a return, and a small op cost amount, then there is no need for fluctuating monthly bills. And if the Kwh are free, then losses don’t matter and higher voltages are not necessary. We freely admit that this is entirely contrary to all electric utility theology, to which we have adhered for lo, these many years.
- The brains of the system. This will be housed in the battery enclosure and is a “battery control system” on steroids. All generated power from any source will come into the battery, and all dispatched power provided to customers will leave via the battery and its inverters. Hence, the managing software must be able to monitor the source(s) of power feeding the system, sense the demand, adjust the supply, and keep records of the whole thing. This is the same set of functions that the Enerblu battery tested for remote military operations is able to provide, although in that case there is a single “customer,” not several hundred of them strung out in a village.
Estimating the cost of this system is difficult, which should not be a surprise since it’s never been built. But there are some basics:
Installed cost of a 5 MW system, ground mounted, no tracking, standard outdoor inverters, should not exceed $1 per watt and could be cheaper. This was the price targeted by the DOE “Sunshot” program, which the government recently announced had been achieved, three years early by the way.
The battery system as specified should cost $600k per Mwh. The control system is in the neighborhood of $1.25 million.
A delivery system serving one 2MW industrial/workshop installation, and residential customers peaking at 300 watts (a light bulb, a fan, a cell phone charger, maybe a small cooker) with power available 24/7 means you can have 1000 rural customers on this system, with a power supply of 5 Mw.
Building out a 1000 customer system is a question of distances. In Florida, a 69Kv line is quoted as $285,000 per line mile. In Africa, we assume lower voltages and lower land, labor and some raw materials costs—wood poles for example. Costs have been cited as $5000 per Km ($8100 per mile), and connection costs as $100 per household. Adding these together at 100 customers per line mile results in an estimated cost of $108,100 per line mile, or $1.081 million total. With a density of 100 customers per line mile, that leaves $6600 per [check this] customer to be amortized by the system.
Capital costs can then be built up as follows:
5 MW solar installation $5 million
5 MWH battery bank $3 million
C&C system $1.25 million
750 Kw Generator $0.75 million
1000 customer distribution system $1 million
Total $11 million
If this amount were all to be amortized as 3% mortgage style debt over 30 years, then the cost per residential customer would be $27.83 per month, assuming that the industrial customer would bear 40% of the expense. Looked at another way, if you assume a system capacity factor of 40%, then the 3MW of residential load will consume 10,512,000 Kwh per year. This is 876,000 Kwh per month for all the residential class, but 876 kwh per month per customer. At this consumption level, cost would be 3.18 cents per Kwh, but with no allowance for operating costs, which in a solar system are very low. Add in a financing structure with an equity return, and some contingency in the capital and you are still at six or seven cents per Kwh.
It is perhaps useful to note that the cost of diesel generated power delivered to these same customers in the bush would probably be eight to ten times this cost.
If some of these costs were different, then of course the results would vary. But delivering under ten cent reliable, non-polluting electricity to rural, unconnected customers in Africa would seem to be an attractive opportunity, with significant ancillary benefits for the customers so connected.
By Robert Hemphill and Leah Bissonette