A Data-Driven Method for Refrigeration Subsystem Performance Mapping with Mode-Based Equation Selection

2026-28-0849

To be published on 11/25/2026

Features
Event
Authors
Abstract
Content
This paper presents a data-driven, mode-based regression method for the refrigeration subsystem of a battery electric truck, in which heating, ventilation and air conditioning (HVAC) cooling, energy storage system (ESS) cooling and simultaneous cooling are treated as five control-determined operating modes. Each mode selects a low-order algebraic equation with its own fitted coefficients to predict compressor work, evaporator, chiller and internal condenser heat rates, and compressor discharge pressure. The equations combine polynomial terms in scaled temperatures with a power term in compressor speed and exponential terms in the flow rates, and an algebraic transient multiplier, driven by the rates of change of compressor speed and flow rates, accounts for compressor cycling in the combined cooling mode without adding states. The coefficients were identified on 16 drive cycles and the model was assessed on 6 held-out drive cycles. On the held-out cycles the coefficient of determination R2 lies between 0.927 and 0.978, the coefficient of variation of the root-mean-square error, CV(RMSE), between 4.3 % and 12.4 %, and the cumulative energy consumption (CEC) is within 5 % of the measurement in five of the six cycles, whereas a conventional polynomial form is within 10 % in only two. The model is algebraic apart from one delay element and suits quasi-steady and transient drive-cycle analysis in a fast-running simulation platform. The evaporator heat rate is the least accurate output.
Meta TagsDetails
Citation
Rajendran, S., Borado, P., Garg, R., and Paranjape, P., "A Data-Driven Method for Refrigeration Subsystem Performance Mapping with Mode-Based Equation Selection," 9th SAENIS TTTMS International Conference 2026, Roorkee, India, December 3, 2026, .
Additional Details
Publisher
Published
To be published on Nov 25, 2026
Product Code
2026-28-0849
Content Type
Technical Paper
Language
English