The UNEP/GEF en.lighten off-grid lighting market model estimates the fuel-based off-grid lighting energy consumption and savings potential from switching to solar LED systems on a one-for-one replacement basis. The model attempts to quantify the savings potential and benefits for 80 countries around the world as their markets leap-frog from fuel-based lighting to efficient solar-LED solutions. The model is based on electrification rates and population demographics, and assigns off-grid technology mixes on a per-person basis based on a small collection of in-depth country studies.
Following is more detail on the methodology utilised in Version 1.0 of the energy-use estimates. Reviewers are invited to comment below with suggestions for improvement, including supporting documentation (reports, spreadsheets, links, etc.) wherever possible.
The countries included in this first model are those with electrification rates less than 90% and a total population greater than 500,000. In addition, country reports were developed for Brazil, China, Mexico, the Philippines, Ukraine and Vietnam because although these countries have electrification rates higher than 90%, they also have a large population base so there are a large number of off-grid households.
The key inputs to the model are discussed below:
Population data - the population estimates are for the year 2010 and are taken from the World Bank Data Catalog, visited October 2012.
Household size - The average household size (i.e., number of people per household) is estimated from data published by the Worldmapper Project, a collaborative effort of the University of Sheffield and the University of Michigan. These estimates are used to determine the number of households that are off-grid. It enables the easy conversion between the units of per person and per household, and for assessment of the 'reasonableness' of resultant estimates.
Electrification rate - The electrification rates are for 2010 and are taken from the World Bank Data Catalog, visited October 2012 and the International Energy Agency's World Energy Outlook 2010. These data are supplemented with recent, reliable input from industry and national surveys. The electrification rate is used to determine whether a country is included in the study and (with the population data) to ascertain the number of people living off-grid in a particular country.
Kerosene prices - the kerosene prices used in the model are meant to represent national average end-user prices. For government subsidized kerosene (as in India), the subsidy portion is not accounted for in the model or the financial benefits calculation, however it will be calculated and reported in the revised version of the model (v.2). Kerosene prices are gathered from numerous sources, including the Lighting Asia Report: Bottom of the Pyramid, the GTZ 2009 Regional Report, UNEP internal sources and national surveys. For countries where there no data is available on kerosene prices, a universal price of US$1.00 per litre is used. The model also considers a high-kerosene price scenario, meant to be representative of rural markets or others with distribution challenges that can increase price. For this sensitivity, a 35% increase over the reference price is applied. Results are therefore presented as a range, from the reference price to the high-price sensitivity.
Operating hours - the model estimates daily average operating hours for the lamps being used in domestic and small commercial applications. For the domestic sector (i.e., households), it is assumed that the average operating hours varies with the number of light points per household. As more light points become available, each light point on average will have lower operating hours. The range of operating hours was established after reviewing studies by Lighting Africa and other experts / organisations, quantifying the operating hours per day. The shape of the curve is defined by an S-curve function, with around 4.4 hours/day for households with one lamp and 3.0 hours/day for households with ten lamps. The shape of the curve is given below, and the operating hours used for each country are presented in the country reports.
For the commercial sector, it is assumed that lamps operate 4 hours per day.
Lighting equipment - due to a lack of country-level data regarding critical inputs such as the lighting technology mix and luminaire prices, the countries modeled have been clustered into five country groups, based around per capita purchasing power parity adjusted gross domestic product (PPP-GDP), and numbers of off-grid population for the five countries that has data available. The clusters of countries were then revised based on any available country-level data (e.g., learning that candles are in common use). The table below presents lighting equipment mix percentages assigned to each of the five groups. The source for these estimates is Lighting Africa.
Light points per person (consumer segment only) - the number of light points per person living in the household for the five African countries is calculated using data from Lighting Africa study. The lamps per room is first calculated using the average lamps per room of a specified technology and the share of the technology of the household lighting. Rooms per household and person per household are then used to get the lighting points per person. The resultant number of light points per person is given for each of the five groups: Ethiopia 0.56 lamps/person; Ghana 0.88 lamps/person; Kenya 0.70 lamps/person; Tanzania 0.65 lamps/person; and Zambia 0.85 lamps/person.
Lighting equipment prices - estimates of the end-user retail prices (in USD) for lighting equipment for seven different off-grid lighting technologies were prepared, based on Lighting Africa data from the detailed studies of five African countries (Ethiopia, Ghana, Kenya, Tanzania and Zambia) as well as other sources. These prices are only used for the pay-back period analysis and are not part of the total cost of lighting services calculation. The off-grid technologies are: kerosene with glass cover, kerosene with a simple wick, torch, a battery-powered light bulb on a wire, candles, and two solar LED lanterns - small and large. All of these prices are presented in the country reports, and are meant to represent national average retail prices.
Note that biomass fuels are currently not accounted for, but will be in revised edition of the model.
Fractional displacement of fuel by off-grid electric lighting - the model works by calculating the current consumption of kerosene, batteries and candles at a household level, and then substituting a solar LED lamp to replace each of those light points. In calculating costs, payback periods and CO2 reductions, this model is a one-for-one replacement assumption, and does not take into account the possibility of the fuel-based light source continuing to be used elsewhere in the household (note: where data are available on this effect, it would be most welcome). Thus, the savings estimates should be regarded as a technical potential, contingent on the efficacy of the underlying technologies as well as policies and deployment programmes.
Data gaps - the model does not take into account any of the public health costs associated with fuel-based lighting, such as poisonings, burns, explosions, house-fires, respiratory disease and so on. Any data or estimates on the frequency of this on a household basis per year or a fuel consumption basis per annum would be welcome. In addition, the model calculates the CO2 savings potential as an indicator of the benefits mitigating climate change, but the model does not yet incorporate the heat-adsorbing effect of black carbon (i.e., soot), which will be included in the revised edition of the model (v.2).
References used for this model
It seems that the assumption about 35% price differential for rural (remote) areas versus baseline should be adjusted per details in the original source. The 35% number is closer to the median across the 5 countries surveyed, rather than an estimate weighted by off-grid population. (The straight average, incidentally, is 60%, but that's because of the Ghana outlier).
I put together the attached spreadsheet, which suggests that the offgrid-population-weighted average is 46%.
We would welcome input from other members on other considerations for adjustments to the official national prices, or those gathered in particular surveys. We would also welcome data on country-specific prices (and urban/rural differentials) for the countries in the study.
Attaching a general reference, where Table 1 (page 4) reports monthly kerosene use for lighting for a list of countries: (Argentina, Benin, Bolivia, Burkina Faso, Honduras, Indonesia, Peru, Sri Lanka, Zimbabwe, and Togo).
I recommend that the "base" scenario excludes all fuel subsidies from the assumed kerosene prices. I've attached a couple of background items. India will be a biggie here.
A "sensitivity" analysis with subsidies will be an instructive way to illustrate how much of the off-grid lighting bill is shouldered by governments, and, by association, all taxpayers.
Some useful data are tucked into the responses to this post.
This is a fabulous resource - thank you! I have a few questions about some of the inputs, listed out below. Having a better understanding on these points would really help d.light to fully and properly utilize this work - as well as improve our own metrics calculations. Thanks!
a) household size: Would you be able to share country-specific inputs you have used?
b) operating hours: I am not entirely clear on how this is being calculated. Is it that the starting point assumption is 4.4 hours for a single lamp, and so any country-level figure below 4.4 suggests households use on average more than 1 lamp (and the inverse for figures that are higher)? Or something different? I would also be very interested in any more detailed explanation that can be provided on how the range of operating hours was established - since this is such a tricky figure to deduce. This section also refers to "lamps" which makes me wonder whether he assumption shifts if we are talking about home systems with multiple LED points instead? And lastly, what is the logic behind the assumption that commercial use lamps operate 4 hours per day?
c) light points per person: Could you clarify where this input is used? And is it applicable only for the 5 countries mentioned?
Dear Darin - thanks for your questions, and sincere apologies for my delay getting back to you on this.
Regarding household sizes - yes, no problem to share the country-specific inputs we used. Our source here was a website called "World Mapper", which is supported by the University of Sheffield (UK), the University of Michigan (US) and a few others: http://www.worldmapper.org/display.php?selected=191. Their sources includes national census data, and the most recent estimate they have published of number of households and number of people per household is 2002. We used these values (assuming household size has not changed in the last ten years).
Regarding operating hours - there is very little data, particularly across the number of countries we were evaluating. However, those studies that do have some discussion on operating hours generally put them between 3 and 6 hours, with fewer average operating hours in households that have multiple light points. Have you seen the 'methodology' report, which provides the equation and plot of the assumed operating hours on pages 3 and 4? LINK HERE Sorry about the term 'lamps' - the point is actually about number of light points, and that average use across all the light points will decline as a house acquires more light points. The logic behind the commercial use was simply that the commercial hours will not scale with number of light points - if a small business has ten lamps or one lamp, they will be used to illuminate things for sale after dark. Small businesses were found in some studies to operate after dark in order to catch those customers returning home from work, thus we assumed four hours per day to cover the time period of 6pm to 10pm, with the same operating hours for all light points, irrespective of the number they own.
Regarding light points per person, this isn't used anywhere as an input to the model. It is a quality control check, calculated on the results in order to identify any countries where there could be a calculation problem or faulty input variable that needs checking. In mature (on-grid) lighting households like the USA, a family has multiple fixed light points per occupant (I would expect >10 light points per person). In emerging (off-grid) lighting markets, there is generally pent-up demand for lighting, or put another way, under-service of illumination. There aren't enough light points per person, and so if a family is able to afford an additional light source, they will do so - but then operating costs are high, so the operating hours should be less (see item above). I added the light points per person check in the model to make sure we didn't calculate excessively high numbers of light points per person. So it isn't an input to the model and it isn't used for anything apart from that quality check.
I hope this helps to clarify matters. Please let me know if you have further questions. Best regards, Mike.
Here's a rather detailed report about kerosene (and solar) among traders in Ghana.
Account for the global-warming potential of black carbon (kerosene as well as candles) ... not to mention biofuels for lighting.
Add reference to exchange rates used....
Bonanza of country-specific kerosene prices here:
But only thru 2008/9. Should check with IMF to see if they've since updated this.
In-depth background data for 5 countries.
Just noticed a table (Appendix 8, page 94) in the latest Lighting Africa Market Trends Report (2012) showing average hours of power outages by country. Perhaps this could be used to develop a better estimate of the "hidden" kerosene and other fuels by nominally grid-connected people.
Here's another amazing source.