Time-series vs typical meteorological year data: verification of PV string sizing and design

Authors: Sevim Zeynep Celik, Marta Pelfort Ojer, Branislav Schnierer, Jozef Rusnak, Giridaran Srinivasan	Presented at IEEE PVSC 2025 in Montreal, Canada
	13.6.2025

Read full document

Overview

Reliable PV system design requires high-resolution, accurate weather data to ensure operational safety and optimal energy yield. This study examines and compares two types of solar weather data commonly used in PV design: Typical Meteorological Year (TMY) datasets and detailed multi-year Time Series (TS) datasets, both from Solargis. Using case studies—including real-world events like 2021's Storm Uri in Texas—the research highlights how TS data captures extreme weather events and high-frequency variability, which are often missing in TMY data but critical for accurate PV string sizing and robust plant operation.

The analysis demonstrates that PV systems designed solely with TMY data are at risk of failing to account for extremes, potentially resulting in system shutdowns, lost energy, and financial losses. Supplementing design with long-term TS datasets allows for a fuller evaluation of operational voltage boundaries and improved reliability across both cold and hot conditions.

Key elements of the study

Data sources

TMY P50 (Typical Meteorological Year): High-resolution (15-minute), synthesized from long-term averages, representing typical or conservative climate conditions; misses year-to-year extremes.
TS (Time Series): High-resolution (15-minute), multi-year records reflecting actual observed variability and rare weather events.

Both datasets are generated by Solargis via satellite-based modeling, with consistent parameters: Global Horizontal Irradiance (GHI), Direct Normal Irradiation (DNI), diffuse irradiation (DIF), and temperature (TEMP).

Design and simulation methodology

Simulations conducted in Solargis Evaluate tool.
Accurate modeling of PV module temperature considering irradiance, air temperature, and module thermal properties.
Modeling of module electrical behavior at the cell, submodule, module, string, and array level, including the effect of bypass diodes and maximum power point tracking (MPPT).
Use of the Martin-Ruiz model for incidence-angle modifier (IAM) losses, and application of air mass (AM) and precipitable water (PWAT) spectral corrections.
String sizing calculations incorporate manufacturer voltage-temperature coefficients and account for extreme cold and hot events affecting module voltage.

Main findings from case studies

Extreme cold scenario: Winter Storm Uri (Texas, 2021)

TS data recorded a rare minimum daytime temperature of –17.1 °C (TMY only captured –1.7 °C).
During this cold spell, simulated PV string voltages exceeded the inverter's 1500 V maximum DC limit for certain string sizes, as flagged in TS-based modeling but missed by TMY.
Actual string sizing risk: With 28 modules per string, TS-based simulation showed over-voltage shutdowns, but TMY did not—demonstrating the danger of relying on simplified/averaged datasets.
Over-voltage episodes can cause system outages, inverter damage, and significant financial losses (e.g., during Storm Uri, Texas plants faced millions in lost revenue due to grid events and repair costs).

Extreme high scenario: Summer heatwaves

High temperatures can cause module voltage to drop below the inverter's minimum MPPT threshold, leading to undervoltage shutdown.
TS data captured more frequent and realistic undervoltage periods than TMY.
For safe operation, longer strings (e.g., 23–24 modules) are recommended during high heat, as shorter strings show more frequent undervoltage events in TS analysis.
TMY underreports the risk and frequency of low-voltage outages compared to TS.

Energy yield and operational reliability

TS-based simulations evaluated string size configurations relative to three references (25, 26, 27 modules) and determined energy gains or losses.
25-module strings were identified as an optimal, safe configuration for cold extremes; longer strings (28, 29) increased shutdown risk without meaningful energy yield benefit.
In heat, configurations with at least 23–24 modules provided better performance and avoided excessive undervoltage episodes.

Conclusion and recommendations

In summary, accurate PV string sizing critically depends on the use of high-resolution time-series weather data rather than TMY datasets alone. Time-series data captures temperature extremes that can lead to overvoltage or undervoltage conditions, reducing system downtime and financial losses. Designs based only on TMY data risk overlooking these operational events, whereas simulations using full time-series datasets result in more reliable and robust PV system performance.