Decentralized Energy and Atmosphere Lab

Perovskite solar cells are an emerging technology with promise for decentralized energy applications. However, these panel designs are still being actively developed. Researchers are tuning device parameters, such as absorber thickness and bandgap to optimize cell performance, typically under standard test conditions (AM1.5G). However, in regions with high aerosol loading, such as the Sahel during Harmattan season, the solar spectrum reaching a PV cell can differ substantially from standard conditions. This is due to wavelength-dependent scattering and absorption by airborne dust and air pollution. Since perovskite cells have different spectral responses than conventional silicon, there is an opportunity to understand how cell design choices interact with real spectral environments. This project develops computational modeling tools that connect local aerosol conditions to cell-level energy conversion physics. Once completed, these models can support the design of solar energy technologies that are robust to the dry and dusty climates common across the African continent. 

Collaborators: Ismaila Dabo (CMU, Materials Science and Engineering)

Decentralized solar systems that serve critical loads like hospital cold storage must be designed to perform reliably under local environmental conditions. In dusty regions such as the Sahel, dust soiling, dust aerosols, and temperature effects can all substantially reduce PV system output, leading to undersized systems that fail to meet demand. Designing systems that are robust to these conditions requires models that can predict performance across a wide range of environmental and design parameters, including panel configuration, battery storage capacity, and local climate. However, the large parameter space makes brute-force simulation computationally expensive. This project develops validated physics-based PV+storage system models that can account for losses from local environmental conditions, and uses them to train machine learning surrogate models that enable rapid performance estimation across the design space. The resulting tools will support the sizing and design of reliable decentralized solar systems in environments where standard assumptions lead to systematic underperformance.

Dust soiling is a persistent challenge for solar panels in arid and semi-arid regions, reducing power output and reliability. For decentralized systems in remote areas, regular manual cleaning is often impractical. Shape-Enhanced Aerodynamic Dust Removal (SEADR) is a passive approach that modifies solar panel geometry to harness wind for dust removal. Preliminary computational fluid dynamics modeling and wind tunnel testing have shown that simple geometrical changes can more than double wind-driven dust removal. This project advances SEADR through continued wind tunnel experimentation, the development of computer vision and computational tools to quantify aerodynamic dust resuspension, and field testing in collaboration with partners in West Africa. Ultimately, we aim to develop panel designs suited to local environmental conditions that can be refined and validated with technicians, installers, and system users in the communities where they will be deployed.

Collaborators: Wesley L. Harris (MIT)