Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Effects of charging and discharging capabilities on trade-offs between model accuracy and computational efficiency in pumped thermal electricity storage

Taemin Heo, Ruaridh Macdonald

Abstract

The increasing need for energy storage solutions to balance variable renewable energy sources has highlighted the potential of Pumped Thermal Electricity Storage (PTES). In this paper, we investigate the trade-offs between model accuracy and computational efficiency in PTES systems. We evaluate a range of PTES models, from physically detailed to simplified variants, focusing on their non-linear charging and discharging capabilities. Our results show that while detailed models provide the most accurate representation of PTES operation by considering mass flow rate ($\dot{m}$) and state of charge (SoC) dependencies, they come at the cost of increased computational complexity. In contrast, simplified models tend to produce overly optimistic predictions by disregarding capability constraints. Other approximated model variants offer a practical compromise, balancing computational efficiency with acceptable accuracy. In particular, models that disregard $\dot{m}$-dependency and approximate nonlinear SoC-dependency with a piecewise linear function achieve similar accuracy to more detailed models but with significantly faster computation times. Our findings offer guidance to modelers in selecting the appropriate PTES representation for their investment models.

Effects of charging and discharging capabilities on trade-offs between model accuracy and computational efficiency in pumped thermal electricity storage

Abstract

The increasing need for energy storage solutions to balance variable renewable energy sources has highlighted the potential of Pumped Thermal Electricity Storage (PTES). In this paper, we investigate the trade-offs between model accuracy and computational efficiency in PTES systems. We evaluate a range of PTES models, from physically detailed to simplified variants, focusing on their non-linear charging and discharging capabilities. Our results show that while detailed models provide the most accurate representation of PTES operation by considering mass flow rate () and state of charge (SoC) dependencies, they come at the cost of increased computational complexity. In contrast, simplified models tend to produce overly optimistic predictions by disregarding capability constraints. Other approximated model variants offer a practical compromise, balancing computational efficiency with acceptable accuracy. In particular, models that disregard -dependency and approximate nonlinear SoC-dependency with a piecewise linear function achieve similar accuracy to more detailed models but with significantly faster computation times. Our findings offer guidance to modelers in selecting the appropriate PTES representation for their investment models.

Paper Structure

This paper contains 27 sections, 16 equations, 42 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Method
  3. An overview of all models
  4. Calculating the PTES capability functions
  5. Model formulations
  6. Model A
  7. Model B
  8. Model C
  9. Model D
  10. Model E
  11. Price-taker case study
  12. Capacity expansion case study using GenX
  13. Results
  14. Price-taker case study
  15. Comparison of model operations
  16. ...and 12 more sections

Figures (42)

  • Figure 1: Conceptual overview of the models from most physically accurate (Model A) to the simplest and common (Model E).
  • Figure 2: Charging and discharging cycles of a PTES system. Operational temperatures are annotated on the flow diagram of the working fluid.
  • Figure 3: Typical charging and discharging capability curves for a PTES system ranging from 30% part-load to full-load conditions.
  • Figure 4: Factored charging capabilities of a PTES system with $\overline{W}_{ch} = 250$ kW illustrating different model formulations.
  • Figure 5: Optimal hourly operations derived from Models A and E for the ERCOT North hub in 2022. The LMP time series is shown in the bottom left panel. The top two panels display the charging and discharging power over the first 100 hours. The two central panels present the SoC time series, and the bottom right panel shows the SoC difference between Models A and E.
  • ...and 37 more figures