Scaling up smart solar home systems in East Africa through improved breakdown of energy consumption, business models and multiplicity of energy use.
Lead Research Organisation:
University College London
Department Name: Civil Environmental and Geomatic Eng
Abstract
In Sub-Saharan Africa, 65% of the population (37% in urban and over 80% in rural areas) live off the grid [1] and have to rely on polluting fuels such as candles, kerosene or wood for lighting and cooking. The traditional energy grid systems have failed to provide energy access to the peri-urban and rural markets. Those markets are currently serviced by a variety of decentralised products, however there is a growing use of financed solar home systems in peri-urban and rural communities across Africa as a replacement for traditional sources of energy. This research will build on the innovative work of BBOXX[2], a London based company who designs, manufactures, distributes and finances innovative plug & play solar systems to improve access to energy across Africa. Since 2014, BBOXX has manufactured and deployed more than 65,000 "smart" (monitored) solar home systems in peri-urban and rural settings of Rwanda. This research will therefore address the following objectives:
Assess how customer usage changes over time and the socio-economic reasons for this change, which could be studied using traditional socio-economic survey methods.
Assess breakdown of consumer usage of energy for different appliances (based on data from 2000 households)
Model future trends in energy consumption considering different appliances usage to assess market demand.
Explore the concept of energy stacking versus the conventional wisdom of consumers progressing through a linear energy ladder.
Evaluate multiplicity of use of energy for domestic and livelihood purposes.
To characterise the relationship between household time-use and energy demand using available data from survey and determine the extent of flexibility in household energy demand
Develop business models for scale up of energy based on current and future energy consumption patterns that maximise flexibility across the energy demand to support energy efficiency and affordability
Assess how customer usage changes over time and the socio-economic reasons for this change, which could be studied using traditional socio-economic survey methods.
Assess breakdown of consumer usage of energy for different appliances (based on data from 2000 households)
Model future trends in energy consumption considering different appliances usage to assess market demand.
Explore the concept of energy stacking versus the conventional wisdom of consumers progressing through a linear energy ladder.
Evaluate multiplicity of use of energy for domestic and livelihood purposes.
To characterise the relationship between household time-use and energy demand using available data from survey and determine the extent of flexibility in household energy demand
Develop business models for scale up of energy based on current and future energy consumption patterns that maximise flexibility across the energy demand to support energy efficiency and affordability
Organisations
People |
ORCID iD |
| Priti Parikh (Primary Supervisor) | |
| Vivien Kizilcec (Student) |
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/N509577/1 | 30/09/2016 | 24/03/2022 | |||
| 1926633 | Studentship | EP/N509577/1 | 30/09/2017 | 29/03/2021 | Vivien Kizilcec |
| Description | My research has progressed the understanding of why customers sign up to solar home systems (SHS), what they use their systems for and what their future energy aspirations will be. This was achieved by fielding surveys in Rwanda with 100 BBOXX customers. A key finding from this research were that the three most important reasons customers signed up for systems, which were clean source of energy, improved and reliable lighting source and charging phones at home respectively. Another key discovery was how much recommendations played a role in solar home system customers' adoption decision. I found that more than half of people recommended SHSs to others, convincing an average of 2 households to also purchase a SHS. A third finding was what type of appliances customers wanted to have, where surprisingly irons were the most sought after new appliance, wanted by 66% of customers. The most common activities that customers use their SHS for are phone charging, security and socialising. I also found out that 76% of customers would like to purchase new BBOXX appliances (upgrade) but most people did not know when or thought it would be more than a year until that time. Finally, this research also found that that a third of customers did not want to be connected to the electricity grid at any point in the future, mainly due to either being satisfied with SHSs or perceiving the grid as a risk to their health. These findings provide an important insight into customers' energy usage patterns. I have also developed a draft convolutional neural network model, which predicts future hourly energy consumption of customers based on their historic energy usage data. I have split the customers into different archetypes based on their socio-demographic characteristics, such as main source of income. The idea is to predict the future energy consumption for each archetype at different intervals, such as daily, weekly and half-yearly. |
| Exploitation Route | Three papers are currently in development and will be submitted this year. The first draft paper is about the outcomes of the fieldwork and is to be submitted to Energy Research & Social Sciences. This will enable other researchers to learn from and build on these findings. As the fieldwork had a relatively small sample size, further studies examining the adoption decisions of solar home system customers would be useful to see whether similar results can be observed. The second paper is a literature review on solar home systems in Sub-Saharan Africa will also soon be submitted to Renewable and Sustainable Energy Reviews. The aim is to classify the papers into four themes of institutional, technological, financial and consumer-centric to identify the current gaps in research, as well as highlight emerging topics within the solar home systems field. This paper will therefore hopefully inspire more research to be conducted in the research gaps highlighted. The third paper will be a conference paper, which was accepted to the RAI2020: Anthropology and Geography: Dialogues Past, Present and Future conference. This paper will be a comparison of the socio-cultural adoption determinants of solar home systems and LPG in Sub-Saharan Africa. The similarities and differences between drivers behind both clean cooking and clean electricity adoption that will be discussed in this paper will provide key information for companies' investment strategies, product design, consumer support and marketing strategies. The machine learning model will be useful for BBOXX because they can ask a household to fill out a quick form when they sign up, which then identifies what archetype they are. Based on this, BBOXX would now know what the typical energy usage profile of that customer would be in the next year and could then recommend the best system capacity for that customer. At the moment, customers are often overpaying for large capacity systems that they do not fully use. This system might enable more household to access energy, as they could pay for cheaper lower capacity systems without worrying that they will run out of energy. |
| Sectors | Communities and Social Services/Policy Energy |
| Description | My fieldwork research asked customers what additional appliances they would like to have in the future to accompany their solar home system. An iron was the most common response and BBOXX will now be offering irons to their customers. |
| First Year Of Impact | 2020 |
| Sector | Energy |
| Impact Types | Societal |
| Title | Predictive Energy Consumption Model |
| Description | I have developed a draft version of a convolutional neural network model, which predicts future hourly energy consumption of customers based on their previous two years of energy usage. I have split the customers into different archetypes based on their socio-demographic characteristics, such as main source of income and their appliances. The idea is to predict the future energy consumption for each archetype at different intervals, such as daily, weekly and half-yearly. Moreover, the model can highlight the regional differences in future energy consumption in Rwanda and Kenya, which may help identify areas of high demand. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2020 |
| Provided To Others? | No |
| Impact | A paper on the model and its uses will be published within the next year. This is still a very new area and almost no work has been done on machine learning models in the rural solar energy sector. This paper will hopefully lead to more research in this exciting area. The machine learning model will be useful for BBOXX because they can ask a potential customer to fill out a quick form when they sign up, which then identifies what archetype they are. Based on this, BBOXX would now know what the typical energy usage profile of that customer would be in the next year and could then recommend the best system capacity for that customer. At the moment, customers are often overpaying for large capacity systems that they do not fully use. This system might enable more households to access energy, as they could pay for cheaper lower capacity systems without worrying that they will run out of energy. |
| Description | 3 Minute Thesis |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Postgraduate students |
| Results and Impact | I held a 3 Minute thesis presentation about the PhD topic in front of an audience of around 50 people. Afterwards, people approached me to learn more about the topic. |
| Year(s) Of Engagement Activity | 2020 |
| Description | Computational Model Showcase |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | I was selected to present my model as part of the Computational Model Showcase in the 3rd eResearch Domain Symposium. I presented my work to around 80 researchers and professional practitioners and received a lot of interest in the model afterwards. |
| Year(s) Of Engagement Activity | 2019 |
| URL | https://www.eventbrite.co.uk/e/3rd-eresearch-domain-symposium-computational-sciences-for-the-21st-ce... |
| Description | Engager training school |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Postgraduate students |
| Results and Impact | The training school brought together international postgraduate students working on data for energy access topics and brought in professional practitioners to discuss new methods and key topics in more detail. Everyone was able to present their projects briefly and a few people were very interested in my work and requested further information. |
| Year(s) Of Engagement Activity | 2019 |
| Description | Talk about Fieldwork |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Postgraduate students |
| Results and Impact | My presentation was about the lessons I learnt from my fieldwork. I provided the audience with tips on conducting their fieldwork, how best to plan it and what mistakes should be avoided. |
| Year(s) Of Engagement Activity | 2020 |