The objective of this work is to develop a flexible, design-based model for the accurate CAPEX estimation of two-tank (hot/cold) Thermal Energy Storage (TES) systems, overcoming traditional "black-box" approaches based on average specific costs or scaling laws that lack transferability across different scales and operating conditions. Given the storage capacity, hot/cold temperatures, fluid thermophysical properties, and geometric constraints, the numerical model performs the physical sizing of the system. It calculates fluid mass and volumes, defines the thickness and mass of the structural steelwork, and estimates insulation, multi-layer foundations, and Balance of Plant (BoP, including piping, heat tracing, and instrumentation) through calibrated correlations, incorporating structural checks for the most critical elements. The output provides a consistent CAPEX breakdown by major categories (tank, insulation, foundation, BoP, and fluid), with the computational chain implemented independently for the hot and cold tanks. Calibration against the baseline dataset yields limited deviations, with maximum errors of 1.1% for the tank, 6% for foundations, and 4% for insulation. The model was subsequently validated against additional literature and industrial references spanning significantly different capacities and temperature levels. For the tank cost, comparison with multi-scale datasets shows a maximum error of 2.8%, confirming the robustness of the sizing procedure and the adopted safety factor. The carbon steel specific cost of 4.4$/kg is validated with some references, while others propose a cost of 5$/kg. For operating temperatures exceeding 450 °C, a multiplier of 2.1 relative to carbon steel specific costs is proposed to account for the use of stainless steel (typically 321H). Regarding insulation, the model exhibits a systematic deviation of approximately 20% in Herrmann cases. This is primarily attributable to discrepancies in specific insulation costs compared to NREL assumptions, indicating that economic sensitivity is dominated by price inputs rather than geometric modelling errors. Foundations represent the most uncertain cost item: besides an outlier (Sandia) with an 83% overestimation due to different subsystem boundaries, other references with itemized breakdowns show deviations between -22% and +20%, consistent with differences in stratigraphy. For aggregated structural items, the typical error remains around 10% for consistent cases. A benchmark case, which validated the model in its low-temperature configuration, exhibited a 27% underprediction in a high-temperature scenario. This discrepancy suggests a potential requirement for an upward adjustment of the stainless-steel correction factor. Sensitivity analysis demonstrates that implementing values up to 3 reduces the deviation to below 5% for the most critical operating conditions.
L’obiettivo della tesi è sviluppare un modello flessibile e design-based per stimare in modo preciso il CAPEX di un sistema TES a due serbatoi (hot/cold), superando approcci “black-box” basati su costi specifici medi o su leggi di scala difficilmente trasferibili tra diverse taglie e condizioni operative. Il modello numerico realizzato, assegnati capacità di accumulo, temperature hot/cold, proprietà termofisiche del fluido e vincoli geometrici, esegue il dimensionamento del sistema: calcola massa e volumi del fluido di accumulo, definisce spessori e masse della carpenteria metallica, stima isolamento, fondazioni multistrato e BoP (piping/heat tracing/strumentazione) mediante correlazioni calibrate, e include verifiche per gli elementi strutturali più sensibili. L’output è un breakdown coerente del CAPEX per macro-voci (serbatoio metallico, isolamento, fondazione, BoP e costo del fluido), con la catena di calcolo implementata separatamente per hot tank e cold tank. La calibrazione sul dataset di base restituisce scostamenti contenuti sulle singole voci di costo, con errore massimo pari a 1.1% sul costo del serbatoio metallico, 6% sul costo delle fondazioni e 4% sul costo dell’isolamento. Successivamente il modello è stato validato rispetto a ulteriori riferimenti bibliografici e industriali, che spaziano tra capacità e livelli di temperatura sensibilmente differenti. Per la voce tank cost, il confronto con dataset multi-taglia (Herrmann) mostra un errore massimo pari a 2.8%, confermando la robustezza della procedura di sizing e del safety factor adottato. E’ stato validato con alcune reference il costo specifico del carbon steel di 4.4$/kg, mentre altre riportano un costo di 5$/kg. Per temperature superiori a 450°C si propone un fattore moltiplicativo di 2.1 al costo specifico del carbon steel, per considerare l’utilizzo di un tank in stainless steel, tipicamente 321H. Per l’isolamento, a parità di geometria il modello mostra uno scostamento sistematico di circa 19% nei casi Herrmann, attribuibile principalmente a differenze nei costi specifici dell’insulation rispetto all’assunzione NREL, ad indicare che la sensibilità economica è dominata dagli input di prezzo più che da errori geometrici. Le fondazioni risultano la macro-voce più incerta: oltre a un outlier (Sandia) con sovrastima pari a 83%, attribuita ad un diverso confine del sottosistema, le altre reference con breakdown separato mostrano scostamenti compresi tra −22% e +20%, coerenti con differenze di stratigrafia. Sulle voci strutturali aggregate, l’errore tipico resta dell’ordine del 10% nei casi più coerenti. Una reference, che ha validato il modello nella configurazione a bassa temperatura, è stata sottostimata del 27% in un caso ad alta teperatura. Ciò suggerisce un possibile aumento del fattore di correzione dello stainless steel. L’analisi mostra che valori fino a 3 consentono di ridurre lo scostamento sotto il 5% nei casi più severi.
Numerical modelling for the design and cost estimation of large capacity thermal energy storage
Venturi, Gianluca
2025/2026
Abstract
The objective of this work is to develop a flexible, design-based model for the accurate CAPEX estimation of two-tank (hot/cold) Thermal Energy Storage (TES) systems, overcoming traditional "black-box" approaches based on average specific costs or scaling laws that lack transferability across different scales and operating conditions. Given the storage capacity, hot/cold temperatures, fluid thermophysical properties, and geometric constraints, the numerical model performs the physical sizing of the system. It calculates fluid mass and volumes, defines the thickness and mass of the structural steelwork, and estimates insulation, multi-layer foundations, and Balance of Plant (BoP, including piping, heat tracing, and instrumentation) through calibrated correlations, incorporating structural checks for the most critical elements. The output provides a consistent CAPEX breakdown by major categories (tank, insulation, foundation, BoP, and fluid), with the computational chain implemented independently for the hot and cold tanks. Calibration against the baseline dataset yields limited deviations, with maximum errors of 1.1% for the tank, 6% for foundations, and 4% for insulation. The model was subsequently validated against additional literature and industrial references spanning significantly different capacities and temperature levels. For the tank cost, comparison with multi-scale datasets shows a maximum error of 2.8%, confirming the robustness of the sizing procedure and the adopted safety factor. The carbon steel specific cost of 4.4$/kg is validated with some references, while others propose a cost of 5$/kg. For operating temperatures exceeding 450 °C, a multiplier of 2.1 relative to carbon steel specific costs is proposed to account for the use of stainless steel (typically 321H). Regarding insulation, the model exhibits a systematic deviation of approximately 20% in Herrmann cases. This is primarily attributable to discrepancies in specific insulation costs compared to NREL assumptions, indicating that economic sensitivity is dominated by price inputs rather than geometric modelling errors. Foundations represent the most uncertain cost item: besides an outlier (Sandia) with an 83% overestimation due to different subsystem boundaries, other references with itemized breakdowns show deviations between -22% and +20%, consistent with differences in stratigraphy. For aggregated structural items, the typical error remains around 10% for consistent cases. A benchmark case, which validated the model in its low-temperature configuration, exhibited a 27% underprediction in a high-temperature scenario. This discrepancy suggests a potential requirement for an upward adjustment of the stainless-steel correction factor. Sensitivity analysis demonstrates that implementing values up to 3 reduces the deviation to below 5% for the most critical operating conditions.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi Venturi.pdf
non accessibile
Dimensione
1.56 MB
Formato
Adobe PDF
|
1.56 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/250961