By Margaret Odero, Data Analyst at nLine
![Figure 1: With the field team at the REMIT Kenya offices in Busia just before we set off for field work. [From left: Ken, Sharon, Margaret (myself), Jane, and Joseph].](/_next/image?url=https%3A%2F%2Fwww.notion.so%2Fimage%2Fhttps%253A%252F%252Fprod-files-secure.s3.us-west-2.amazonaws.com%252F982860da-8ce1-44e1-b3cd-b7c663782bba%252F7c9878fc-ff49-4ebd-94d5-27e381bf16bb%252Fkindle.png%253FspaceId%253D982860da-8ce1-44e1-b3cd-b7c663782bba%3Ftable%3Dblock%26id%3D138c318d-7fc8-4908-b4aa-aa632bf3c331%26cache%3Dv2&w=3840&q=75)
Over the past few years, the Government of Kenya (GoK) has made significant efforts towards connecting Kenyan households to the national grid. The progress has been marked by several national projects, one of them being the Last Mile Connectivity Program (LMCP). Efforts such as LMCP have seen Kenya’s electricity access increase to 75% of its population compared to just over 40% in 2015, as per the World Bank records. The LMCP — one of the country’s largest public infrastructure projects (costing roughly $400 million) — was jointly funded by African Development Bank (AfDB) and the World Bank. Each of these funders, as is the case with most multilateral organizations that fund projects in low and middle-income countries, had some level of donor conditional requirements to fulfill during the project.
A team of researchers from UC Berkeley and University of Pennsylvania, working in collaboration with nLine, set out to investigate whether the conditions imposed by the African Development Bank led to different power quality and reliability outcomes than the stricter World Bank conditions. In this project, nLine sensors are being deployed to measure customer-level power quality and reliability at high spatial and temporal resolution. By analyzing data from multiple sensors, nLine produces rich analytics on grid quality and reliability to enable the research team to evaluate whether more stringent donor conditions on infrastructure construction improves power quality on the ground.
In this post, I give an account of my experience when I joined the field team for a week in October 2021 as they deployed sensors across counties in Kenya for this study.
From Baseline Surveys to Sensor Deployments
This study had begun earlier in 2016 with baseline surveys of households in sites across 5 counties in Kenya (Kisumu, Kakamega, Vihiga, Nandi, and Kericho). Field officers collected and geo-stamped data on electricity use and other socioeconomic factors.
The next step was to deploy sensors. Beginning in May 2021, surveyors began returning to baseline study participants’ homes to install nLine sensors. The sensor is plugged directly into wall outlets to monitor voltage and frequency levels as well as duration and frequency of outages. Sensors are installed for 2 months in 25 sites, after which the sensors are collected and redeployed in 25 new sites (note: four sensors are installed per site — one per household connected to the same transformer). For the study, “sites” are defined as LMCP construction locations — homes or business premises — that are connected to the same transformer.
It was at the stage of sensor collection and redeployment that I jumped on the opportunity to have first-hand experience with the steps that go into obtaining the raw data measurements that I have been spending months analyzing and generating insights from.
Deployment 101: Shadowing the Installation and Pickup of Sensors
On October 17th 2021, I traveled to Busia to meet the field team at the REMIT Kenya offices to join them in picking up and re-deploying sensors. By this time, I had never actually held a sensor in my hand, even though it is the source of the data with which I had worked for months. Neither had I interacted with the communities from which this data was collected, and it made me both nervous and excited. I was actually going to be in touch with a majority of the process, if not the whole data collection process.
We met and debriefed that evening in preparation for the work that lay ahead. It was easy to blend in with the field team, and this obviously happened over a nyama choma meal because… what else do you bond over when you are Kenyans 😄.
We set off the next morning with a mission to collect and redeploy sensors from five different counties. The team was split into two groups and I joined the one which would work in Kisumu and Kericho. This would mean visiting a total of 10 deployment sites over the next week.
In each site, sensors were installed in four households, with two households closest to the transformer and two households furthest from the transformer (the idea was to measure any changes in power quality and reliability as a household moved further down the line from a transformer). Many other households connected to the transformer had been interviewed during the baseline survey, and served as backup participants in case it was not feasible to deploy sensors in the four originally identified households.
The process involved driving from county to county, site to site, home to home picking up sensors where they had been deployed and redeploying them in new sites. It was more intense than I was prepared for because of the early mornings and long days we had to work through. The early morning starts — 7am to be precise — were necessary to have enough time for collection and deployment of sensors in different counties.
My role was to observe the workflow of field officers as they explained the purpose of installing the sensors to the participants, plugged the sensors in, and answered any follow-up questions from participants.
Mastering the Art (and Challenges) of Sensor Deployment
Watching the interaction between the field officers and the participants was enlivening. I was pleasantly surprised at how willing participants were to cooperate with the field officers. If they were hesitant at first, the field officers had a charisma that helped participants feel comfortable. Furthermore, having been involved in the initial baseline demographic survey months earlier, participants often recognized the field officers or the organization name and were willing to participate. Some participants were simply grateful that they had benefited from LMCP and had finally been connected to the grid, and therefore were willing to participate in related research.
However, not everything went entirely smoothly during this process. There were some language barriers which led to misunderstanding of the functionality of the sensor. For example, one of the respondents thought the sensor was a backup power source. There were also cases where gender played a role in ease of having respondents participate and male figures of the house had to give consent for sensor installation.
Given the hurdles field officers had to deal with, including cases where participants’ homes were difficult to locate due to lack of precision of the geolocations, I could not help but notice just how patient and stoical they were. It was evident that they had mastered the art of fieldwork.
Closing Thoughts
In retrospect, it was very important for me to get out from behind my computer and see the field activities. Not only did I understand the data process better, but it hugely reignited in me the impact-centered mission of the work I was doing and nLine’s work at large. It was super exhilarating to see how happy people were that LMCP granted them access to electricity. However, seeing the disappointment on their faces when they talked about how frequently the power always went out was heartbreaking. It rekindled empathy and made me realize the urgency of nLine’s work in improving electricity reliability.
To learn more, visit our website and read these related blog posts on engineering lessons learned from this deployment and early insights on customer-level voltage quality.