Modelling, Control and Optimisation of Electric Transport Refrigeration Unit
Authors
Hetherington, Andreas ; Abbas, Ryan Sebastian
Term
4. term
Education
Publication year
2023
Submitted on
2023-06-02
Pages
141
Abstract
Transport af fødevarer, medicin og andre varer gør effektiv køling stadig vigtigere, og elektrificering øger behovet for lavere energiforbrug og højere virkningsgrad. Dette projekt modellerer, styrer og optimerer en elektrisk transportkøleenhed baseret på en flashtank-dampkompressionscyklus. Den centrale problemstilling er at fastsætte og regulere flashtankens mellemtryk og andre setpunkter for at reducere energiforbruget uden at kompromittere køleydelsen. Der udvikles en højopløselig dynamisk model af systemet med komponentniveau-masse-, energi- og trykdynamik samt en tilhørende tilstandsrumbeskrivelse. Der udformes styringsstrategier for superheat, subcooling, reefer-lufttemperatur og flashtanktryk, som vurderes via simulering og sammenlignes med tilgængelige BITZER-data. Til optimering anvendes en genetisk algoritme til at søge efter et optimalt mellemtryk; sværmalgoritmer diskuteres også. Resultaterne peger på, at GA-baseret setpunktssøgning er mulig, når kølemidlets trykdynamik er tilstrækkeligt stabil, men at få optimerbare tilstande i den nuværende model begrænser nytten af gradientfri optimering. Arbejdet peger på yderligere potentiale med flere optimerbare setpunkter og en mere forfinet modellering.
The transport of food, medicine, and other goods makes efficient refrigeration increasingly important, and electrification raises the need for lower energy use and higher efficiency. This project models, controls, and optimizes an electric transport refrigeration unit using a flash-tank vapor-compression cycle. The core question is how to determine and regulate the flash-tank intermediate pressure and other set points to reduce energy consumption without compromising cooling performance. A high-fidelity dynamic model is developed with component-level mass, energy, and pressure dynamics and an associated state-space representation. Control strategies are designed for superheat, subcooling, reefer air temperature, and flash-tank pressure, evaluated through simulation and compared against available BITZER data. For optimization, a genetic algorithm is applied to search for an optimal intermediate pressure; swarm methods are also discussed. The reported results indicate that GA-based set-point search is feasible when refrigerant pressure dynamics are sufficiently stable, but the limited number of optimizable states in the current model reduces the benefit of gradient-free optimization. The work suggests further potential with a broader set of optimizable set points and refined modeling.
[This summary has been generated with the help of AI directly from the project (PDF)]
Keywords
Documents