Creep prediction of cellulose based materials by extrapolation of short term experiments

Durability of cellulose-based insulation in power transformers is a key element to guarantee the reliability of power transmission systems. Under constant loading, continuous deformation is called creep, and its accurate prediction is essential for estimating the service life for constructions under dead load. Conducting decade-long experiments is technically possible yet practically unattainable at the design stage; therefore, we search for reliable methods that use shorter experiments (months) to predict long-term behavior (years). Extrapolation of measurement data often induces significant deviations from the actual response, even when the prediction time is only three to five times longer than the experimental data. Specifically, in this work, we study creep behavior, which is the dominating response in cellulose and related biopolymers for decade long static loading, especially in the presence of moisture. We analyze pressboards used in technical applications that exhibit a relaxation time of several hours, making the choice of an accurate material model challenging. By implementing three different approaches from the literature for modeling the creep behavior, we demonstrate a simple yet effective methodology to predict time spans longer than the experiment itself. The creep response of a precompressed pressboard has been analyzed, and with the presented data-driven method by means of an inverse analysis, a short-time experimental investigation is sufficient to characterize the creep properties, which then predicts the mechanical behavior for time spans five times longer than the experimental data's duration. The study demonstrates this approach under [Formula: see text] RH (Relative Humidity) conditions, the same methodology may be used for different moisture levels in the future.