nLine’s Blog
nLine’s Blog
Using Power Source Detection as Evidence for the Energy Transition
nLine.io

Using Power Source Detection as Evidence for the Energy Transition

Date
Jan 8, 2026
By: Margaret Odero (Lead Data Analyst, nLine) and Olufolahan Osunmuyiwa (Director of Partnerships, nLine)
 
page icon
Renewable and hybrid energy systems are expanding rapidly across sub-Saharan Africa—but proving their real climate impact remains a challenge. This blog outlines why this verification gap matters and shows how nLine uses high-frequency voltage and frequency data to measure when solar is truly replacing diesel, enabling more credible emissions accounting and unlocking pathways to climate finance.
 

1. Renewable Energy is Africa's Tool to Fight Energy Poverty and Climate Change

Renewable energy (RE) sources have become a key tool in Africa’s campaign to eliminate energy poverty and deliver a decarbonized, industrialized, and climate resilient continent. In 2024, Africa’s renewable power capacity increased by 4.2 GW (IRENA, 2025) with solar photovoltaics (PV) contributing approximately 2.4 GW (57.14%) of the recorded growth. Despite accounting for just 1.5% of global renewable capacity, current projections suggest that the region could electrify its nearly 600 million people without electricity (World Economic Forum, 2024) and produce 10 times its electricity needs by 2040 solely from renewables.
Figure 1: Renewable Power Capacity Additions by Region/Country, 2024. Adapted from (REN21, 2025). The total renewable capacity for Africa and Middle East combined was 13GW, less than 2% of the gloabal capacity.
Figure 1: Renewable Power Capacity Additions by Region/Country, 2024. Adapted from (REN21, 2025). The total renewable capacity for Africa and Middle East combined was 13GW, less than 2% of the gloabal capacity.

2. Multiple Actors are Supercharging Renewable Energy Investments—Especially for Health Electrification

Development finance institutions (DFIs), successive Conferences of the Parties (COPs)— and national actors have been instrumental in championing renewable-energy as a catalyst for SSA’s clean electrification, the reduction of fossil-fuel dependency and a means to achieve climate-mitigation objectives, including those related to carbon markets and climate finance. For example, in Sierra Leone solar PV deployments under national energy access programs—such as the Sierra Leone Healthcare Electrification Project— have reached hundreds of health facilities to "replace unreliable grid electricity and polluting diesel generator capacity with renewable energy solutions that can deliver reliable, clean power."
Figure 2: Lakka Government Hospital solar PV system installed under the Sierra Leone Healthcare Electrification Project. This installation makes Lakka Hospital a hybrid system as it is also connected to the grid.
Figure 2: Lakka Government Hospital solar PV system installed under the Sierra Leone Healthcare Electrification Project. This installation makes Lakka Hospital a hybrid system as it is also connected to the grid.

3. Despite Renewable Energy Aspirations, Solar-Diesel Hybridization Persist

Even as RE interventions expand, persistent reliance on diesel remains a struggle for the region. The impact of this challenge is particularly pronounced in health facilities. For instance, more than 30% of "electrified" health facilities in the Gambia still rely exclusively on diesel generators (WHO, 2023), underscoring both the scale of opportunity for renewable energy deployment and the persistence of fossil fuel dependence. Similarly, across East and West Africa, newly electrified health facilities are implementing solar-diesel hybrid systems. Yet, the accounting for continued solar-diesel hybridization does not hold up, because current data shows that electricity generation from solar PV is way cheaper than diesel.

4. Displacing Fossil Fuels Can Lead To Carbon Market Gains

Beyond costs, displacing current fossil fuels through investments in renewables can; (i) help SSA address emerging climate fragility, (ii) enable the region’s participation and earnings in the currently estimated 43.3 billion carbon market and (iii) be a more decolonized path than the current dominant pathway where countries give up “swaths of forest for carbon removal, and in the process, lose sovereignty over their land” (Adesina, 2025).

5. Verifying True Fossil-Fuel Displacement Remains a Challenge

There is an obvious economic and ecological benefit to renewable energy integration across Africa. However, for projects to translate into carbon credits under voluntary or compliance carbon markets, their emission-reduction methodologies must be quantified, traceable, and approved by recognized accreditation bodies such as Gold Standard, Verra, or within frameworks consistent with the Greenhouse Gas Protocol and ISO 14064.
Unfortunately, valuating and verifying the actual scale of fossil fuel displacement achieved by renewable energy projects in SSA has remained a challenge. For example, a roadmap for distributed renewable energy (DRE) in Africa notes:
“There is no meaningful precedent for DRE‑related carbon revenues in Africa. This is largely driven by a lack of awareness and lack of clarity on pathways to certify DRE projects for carbon credit generation.” — Africa Carbon Markets Initiative (ACMI)
Likewise, Kenya’s draft National Energy Policy identifies “lack of adequate capacity on project monitoring and verification” as a key challenge for carbon‑finance and clean‑energy projects.

6. Measurement, Reporting and Verification Gaps Are Opportunities for Technological and Fiscal Innovation

The current gaps in measurement, reporting, and verification (MRV) mean that although many solar or hybrid energy systems are installed, the evidence base for quantifying the extent to which solar generation has replaced diesel usage, the resulting reduction in CO₂ emissions, and the duration for which these systems sustainably offset generator use remains weak. This presents both technical and fiscal opportunities. From a technical standpoint, there is an opportunity for innovative measurement approaches that provide granular time-stamped data on power-source activity. Such data could help determine the duration and proportion in which solar-generated power replaces diesel-generated power even within hybrid systems. This also provides a structured pathway to account for and validate fossil fuel displacement (carbon) claims, provide traceability, and certify integrity. From a fiscal perspective, measurement approaches that enable proper quantification and valuation of fossil fuel displacement ratio and scale can help raise the creditworthiness and improve the risk profile of small renewable energy developers in SSA - which can in turn help unlock substantial capital for their growth.
Below, we describe nLine’s innovative power-source detection approach and explain how this data-driven methodology can quantify the climate and economic impacts of renewable interventions in SSA, support carbon-footprint accounting, and inform future accreditation within carbon-market frameworks.

7. nLine Turns Data Into Verified Impact

nLine’s power-source detection algorithm uses granular frequency measurements to infer whether a site is drawing from grid, solar, or generator supply to create an evidence base for quantifying true solar substitution, fuel savings, and emissions reduction. This algorithm is a lightweight statistical, rule-based system designed for reliability, interpretability, and low computational overhead. To illustrate the efficacy of the algorithm, we use nLine data collected from hybrid grid-connected and off-grid hospitals that were solarized in Sierra Leone. Our sensors were installed in different rooms in these healthcare facilities to collect voltage and frequency data at two-minute intervals.

7. 1. Frequency as a Diagnostic Signal

In alternating-current (AC) systems, frequency serves as a global indicator of supply-demand balance. In a synchronous grid, all rotating machines are mechanically coupled through the electrical network, sharing a common frequency (in Sierra Leone and most of SSA the nominal frequency is 50 Hz). When generation lags demand, frequency falls; during excess supply, it rises. This system-wide coupling gives frequency its diagnostic power. Synchronous generators possess rotational inertia that resists rapid frequency changes, producing relatively slow and correlated fluctuations across all grid-connected sites.
Inverter-based systems, by contrast, establish frequency electronically through control algorithms. Grid-forming inverters tightly regulate their internal oscillators, producing frequency signals that cluster very narrowly around nominal values.
Diesel generators, typically small synchronous machines with mechanical governors, exhibit broader frequency fluctuations due to the combined effects of governor response lag and droop control, which cause frequency to vary proportionally with load changes. Compared to the larger grid-connected synchronous generators, their limited rotational inertia also contributes to faster and more pronounced deviations.
Empirical analysis of nLine’s datasets from Sierra Leone confirmed these theoretical expectations:
  • Grid-supplied hospital rooms display moderate variance around 50 Hz.
  • Solar systems, particularly those using modern inverter controllers, exhibit minimal deviation (± 0.2 Hz).
  • Generators show the widest spread, from 43 Hz to 57 Hz under varying load.
These distinctions, while sometimes not perfectly separable, provide a robust first layer for classification.
Figure 3: Empirical frequency data from four hospital rooms served by: (i) generator, (ii)solar, (iii)grid, and (iv) generator-solar hybrid systems. The frequency signatures for different power sources are distinctive for the most part based on frequency value and variance/stability of the signal.
Figure 3: Empirical frequency data from four hospital rooms served by: (i) generator, (ii)solar, (iii)grid, and (iv) generator-solar hybrid systems. The frequency signatures for different power sources are distinctive for the most part based on frequency value and variance/stability of the signal.

7.2. nLine’s Rule-Based Detection Methodology

The methodology combines nLine’s sensor-reported frequency data with derived features, ground-truth verification, local grid knowledge, and domain knowledge of how different power sources imprint distinct frequency signatures. Supported by statistical analysis, these components enable systematic characterization of the frequency signal and its features, allowing individual two-minute data points to be classified as “solar,” “grid,” or “generator.”
The classification process begins with the raw frequency measurements. Each sensor’s two-minute reading is compared against empirically derived thresholds that capture the distinct precision levels of the three power sources. Solar systems show high precision with frequency measurements tightly clustered around the 50 Hz nominal value. Grid supply exhibits moderate precision, remaining near 50 Hz but with noticeably greater variance than solar. Finally, generator outputs exhibit the lowest precision, with broad deviations from the 50 Hz nominal and observed values ranging from approximately 43 Hz to 57 Hz.
Because absolute frequency values can overlap, particularly between solar and grid, the algorithm incorporates a spatial-reference comparison. For each monitored site, the algorithm identifies the nearest grid-only facility and calculates the deviation between the sensor’s frequency and the median grid reference frequency at the same timestamp. If this deviation is less than the empirically determined threshold, the reading is labeled as “grid”. Sites connected to the same synchronous grid should have the same frequency. This is because in a synchronous AC power system, all generators behave as coupled oscillators and must rotate in synchronism at the same rotor speed and AC fields propagate almost instantaneously across the network. The entire synchronous grid therefore shares one instantaneous frequency.
Temporal stability serves as an additional differentiator. Using a sliding window around each data point, the algorithm computes the mean absolute deviation (MAD) of the sensor-reported frequency. This window differentiates short-term behavior across power sources: solar systems show very low MAD due to their stable inverter output; grid supply exhibits moderate MAD, reflecting small but consistent fluctuations; and generators produce high MAD because of their inherently irregular frequency. A low MAD is therefore a strong indicator of solar operation, while higher values point toward grid or generator behavior.
During nLine's sensor deployment in healthcare facilities, we recorded the power sources supplying each sensor-monitored room. These ground-truth labels of power sources are used to refine and validate the classification rules. Rooms powered exclusively by a single source (solar, grid, or generator) provide unambiguous reference data and are critical for calibrating thresholds, verifying frequency signatures, and resolving classification errors that may arise from general rules.
The resulting classifications are refined iteratively as additional ground data become available. Over time, the labeled datasets—grounded in rigorously validated frequency signatures, spatial-reference comparisons, and temporal stability rules—will serve as highly reliable training material for supervised machine-learning models, allowing nLine to expand from rule-based inference to adaptive, data-driven prediction.
Figure 4: A summary of nLine’s rule-based power source detection methodology. It relies on frequency, feature engineered variables, thresholds from empirical data analysis, knowledge of power sources supplying a facility, a smoothing process, and process improvement based on new information. The resulting labeled data can be used to train supervised machine learning models for a more generalized application.
Figure 4: A summary of nLine’s rule-based power source detection methodology. It relies on frequency, feature engineered variables, thresholds from empirical data analysis, knowledge of power sources supplying a facility, a smoothing process, and process improvement based on new information. The resulting labeled data can be used to train supervised machine learning models for a more generalized application.
 

8. From Power Source Classification to GHG Emissions

8.1 Extent and Duration of Solar Replacement of Diesel

Once each two-minute interval is labeled by power source, these periods are aggregated to determine the total duration of use of each power source for any desired temporal resolution. The results are as shown in Figure 5 below for 4 government hospitals in Sierra Leone.
Figure 5: Solar PV replacement of diesel generators in 4 Government Hopsitals in Sierra Leone. We can quantify the extent and duration of solar replacement through sensor measurements.
Figure 5: Solar PV replacement of diesel generators in 4 Government Hopsitals in Sierra Leone. We can quantify the extent and duration of solar replacement through sensor measurements.

8.2 Quantifying CO₂ Emissions Reduction

With the determined durations of power source usage, carbon estimation becomes a direct arithmetic step. For periods labeled as “generator”, nLine multiplies the time duration by the generator’s rated fuel-consumption rate, obtained from the manufacturer or reference charts such as FW Power, Generator Source, or Able Sales, and by the standardized diesel emission factor of 2.68 kg CO₂ per liter, according to the U.S. Environmental Protection Agency and corroborated by Michelin Connected Fleet and Driver Knowledge Test. Summing these emissions yields monthly or annual estimates of generator-related CO₂ output. Because the underlying run-times are measured rather than assumed, the resulting carbon footprints reflect the actual operational behavior of each facility.
Figure 6: Monthly and total carbon emissions from diesel generatos at 6 hospitals in Sierra Leone for a one year period before and after intervention (solar installation). Carbon emissions reduced by ~82% as a result of solar installations.
Figure 6: Monthly and total carbon emissions from diesel generatos at 6 hospitals in Sierra Leone for a one year period before and after intervention (solar installation). Carbon emissions reduced by ~82% as a result of solar installations.
Equivalent logic can extend to solar and grid intervals using their respective emission factors, enabling full decomposition of a facility’s carbon profile.
This methodology conforms with recognized carbon-accounting frameworks, including the European Investment Bank’s Project Carbon Footprint Methodologies and the European Commission’s Methodological Approach for Defining Emissions from Electricity Consumption. What differentiates nLine’s approach is temporal precision: the capacity to tie emissions directly to observed energy transitions, rather than averaged consumption figures.

9. nLine’s Pathway to Methodological Accreditation

nLine remains committed to continuously improving the above methodology. Our plan is to align our measurement processes and quantification with internationally recognized carbon-accounting standards and to purse accreditation from reputable systems, such as the Greenhouse Gas Protocol, ISO 14064, the Gold Standard, and Verra’s Verified Carbon Standard.
 
Figure 7: An excerpt from the ISO 14064, an internationally recognized standard that helps businesses measure, manage, and report their greenhouse gas (GHG) emissions.
Figure 7: An excerpt from the ISO 14064, an internationally recognized standard that helps businesses measure, manage, and report their greenhouse gas (GHG) emissions.
nLine’s pathway toward accreditation involves three main stages:
  1. Codifying the methodology, documenting data sources, frequency-analysis logic, emission-factor selection, uncertainty treatment, and quality-assurance processes.
  1. Engaging independent auditors to validate performance and reproducibility across diverse sites and geographies.
  1. Seeking recognition from one or more of the above standards, either as a supplemental monitoring method under an existing framework or as a stand-alone methodology for hybrid-energy systems.
Formal accreditation would allow nLine’s outputs to serve not only for operational analytics but also for verified reporting under carbon-credit or sustainability-disclosure programs. It would link real-time energy behavior at hospitals and clinics to globally recognized emissions metrics, closing a persistent data gap in infrastructure carbon accounting.

10. Finally, nLine’s Goal is to Translate Verified Impact into Market Opportunities

As the conversation around climate finance, higher-integrity carbon markets, and just energy transitions continue, our proposed methodology for calculating fossil-fuel displacement gives African renewable-energy projects exactly what they need at this pivotal moment: credible, transparent proof of the emissions they avoid and the real-world impact they deliver. Projects that can demonstrate precise fossil-fuel displacement will be better positioned to attract investment, qualify for premium carbon credits, and participate in the next wave of global climate mechanisms. Communities benefit too—because rigorous measurement helps direct finance toward energy solutions that genuinely improve livelihoods, reliability, and affordability.
nLine is ready to collaborate with project developers, investors, and policymakers to seize this moment. As carbon markets tighten their expectations around additionality, baselines, and monitoring, we can support partners in applying this methodology end-to-end: gathering data, modelling displacement, building fit-for-purpose MRV frameworks, preparing documentation for high-integrity carbon markets, and translating verified impact into bankable opportunities. Our goal is to help African renewable-energy projects gain competitive edge through our transparent, data-driven and verifiable platform.
We welcome collaboration, feedback, and new partnerships; if you would like to engage with this work, please get in touch at info@nline.io.
 

11. References

11.1. Data Sources

  1. Adesina, A. (2025). Carbon grabs threaten Africa’s sovereignty. Yale Environment 360. https://e360.yale.edu/digest/akinwumi-adesina-carbon-grabs
  1. Africa Carbon Markets Initiative. (2022). Roadmap report: Accelerating Africa’s participation in carbon markets. https://energyalliance.org/wp-content/uploads/2022/10/comp_ACMI_Roadmap_Report_Nov_2022.pdf
  1. Able Sales. (n.d.). Diesel generator fuel consumption chart (litres). https://www.ablesales.com.au/blog/diesel-generator-fuel-consumption-chart-in-litres.html
  1. Driver Knowledge Test. (n.d.). Why does burning 1 litre of fuel create over 2 kg of carbon dioxide? https://www.driverknowledgetests.com/resources/why-does-burning-1-litre-of-fuel-create-over-2kg-of-carbon-dioxide/
  1. FW Power. (2018). Diesel generator fuel consumption chart (litres). https://fwpower.co.uk/wp-content/uploads/2018/12/Diesel-Generator-Fuel-Consumption-Chart-in-Litres.pdf
  1. International Renewable Energy Agency. (2025). Renewable capacity statistics 2025. https://www.irena.org
  1. Michelin Connected Fleet. (n.d.). How to calculate CO₂ emissions. https://connectedfleet.michelin.com/blog/calculate-co2-emissions
  1. REN21. (2025). Renewables global status report 2025. https://www.ren21.net
  1. SEforALL. (n.d.). Sierra Leone healthcare electrification project. https://www.seforall.org/programmes/powering-healthcare-hub/sierra-leone-healthcare-electrification-project
  1. World Economic Forum. (2024). How Africa can power a renewable energy future. https://www.weforum.org
  1. World Health Organization. (2023). Energy access in health facilities: The Gambia country profile. https://www.who.int
  1. WWF. (2025). UN energy transition report. https://wwfint.awsassets.panda.org/downloads/final_un-energy-transition-report_2025.pdf

11.2. Policy & Standards

  1. European Commission. (2021). Methodological approach for defining greenhouse gas emissions from electricity consumption. https://ec.europa.eu/regional_policy/sources/policy/evaluations/guidance/2021/ghg-methodologies/renewable-cogeneration.pdf
  1. European Investment Bank. (2023). Project carbon footprint methodologies. https://www.eib.org/attachments/lucalli/eib_project_carbon_footprint_methodologies_2023_en.pdf
  1. Gold Standard. (n.d.). Gold Standard methodologies. https://www.goldstandard.org
  1. Greenhouse Gas Protocol. (n.d.). GHG Protocol standards. https://ghgprotocol.org
  1. ISO. (2018). ISO 14064-1: Greenhouse gases—Part 1: Specification with guidance at the organization level for quantification and reporting of greenhouse gas emissions and removals. https://www.iso.org/standard/66453.html
  1. Kenya Ministry of Energy. (2025). Draft national energy policy 2025–2034. https://energy.go.ke/sites/default/files/Final%20Draft%20NEP%202025-2034%20%281%29.pdf
  1. U.S. Environmental Protection Agency. (n.d.). Greenhouse gas emissions from a typical passenger vehicle. https://www.epa.gov/greenvehicles/greenhouse-gas-emissions-typical-passenger-vehicle
  1. Verra. (n.d.). Verified Carbon Standard methodologies. https://verra.org/methodologies/