Mixing time prediction in stirred tanks using empirical methods and computational fluid dynamics
(2024) KLTM06 20241Food Technology and Nutrition (M.Sc.)
- Abstract
- Stirred tanks are essential components in design and optimization of food processing lines, as they assure proper mixing and homogenization of ingredients. In this context, the mixing time is one of the most impactful performance parameters affecting various process design decisions. Thus, reliable methods to accurately predict mixing times in stirred tanks are highly desired.
The aim of this thesis is to expand Tetra Paks knowledge in mixing time prediction by using empirical methods and computational fluid dynamics (CFD) to predict the mixing time in one of their pilot tanks. Experimental mixing times had been produced prior the thesis, which are used to compare and validate the predictions against. CFD is used by developing a mixing... (More) - Stirred tanks are essential components in design and optimization of food processing lines, as they assure proper mixing and homogenization of ingredients. In this context, the mixing time is one of the most impactful performance parameters affecting various process design decisions. Thus, reliable methods to accurately predict mixing times in stirred tanks are highly desired.
The aim of this thesis is to expand Tetra Paks knowledge in mixing time prediction by using empirical methods and computational fluid dynamics (CFD) to predict the mixing time in one of their pilot tanks. Experimental mixing times had been produced prior the thesis, which are used to compare and validate the predictions against. CFD is used by developing a mixing time prediction method based on recent research that uses mean age theory (MAT).
CFD simulations of the pilot tank were conducted using two different agitators, one multi-impeller setup with baffles (MI setup) and one single-impeller setup without baffles (SI setup). The SM method was used in the simulations where various mean age distributions were extracted and used to obtain mixing time predictions, which were compared with the experimental data and empirical predictions.
For the MI setup, we found that an empirical method performs well in the turbulent and transitional regime. The CFD method also provides accurate predictions in the turbulent regime, and in the transitional regime with some method tweaks. This is however deemed less useful due to the accuracy of the less time consuming empirical method. Neither empirical methods nor CFD were able to predict the mixing time in the laminar regime. For the SI setup, we found that its dissatisfactory mixing performance makes the empirical predictions inadequate, and that the CFD method gave proper predictions with some method tweaks. Generally, we concluded that a CFD
method is valuable in the laminar regime and for the SI setup where empirical correlations lack applicability and performance. However, the tweaks required using the MAT method makes it significantly less robust, why other CFD methods are considered more suitable in comparison to MAT in its current state. (Less) - Popular Abstract (Swedish)
- Flera miljarder dollar går uppskattningsvis förlorade årligen p.g.a. något tämligen väldigt enkelt; omblandning. Det finns alltså oerhörda förbättringsmöjligheter inom detta område i olika industrier, vilket detta arbete handlar om.
Omblandning eller mixning är ett mycket vanligt steg i olika tillverkningsprocesser i livsmedelsindustrin, där ingredienser oftast mixas i en omblandad tank till en homogen blandning såsom yoghurt. Vid design av tillverkningsprocesser är kunskap om mixningsprestandan, framförallt mixningstiden, mycket värdefull, p.g.a. de gigantiska ekonomiska konsekvenserna av ineffektiv mixning. Det är därför eftersträvansvärt med metoder som predikterar mixningstiden i tankar med olika livsmedel.
Detta arbete gick ut... (More) - Flera miljarder dollar går uppskattningsvis förlorade årligen p.g.a. något tämligen väldigt enkelt; omblandning. Det finns alltså oerhörda förbättringsmöjligheter inom detta område i olika industrier, vilket detta arbete handlar om.
Omblandning eller mixning är ett mycket vanligt steg i olika tillverkningsprocesser i livsmedelsindustrin, där ingredienser oftast mixas i en omblandad tank till en homogen blandning såsom yoghurt. Vid design av tillverkningsprocesser är kunskap om mixningsprestandan, framförallt mixningstiden, mycket värdefull, p.g.a. de gigantiska ekonomiska konsekvenserna av ineffektiv mixning. Det är därför eftersträvansvärt med metoder som predikterar mixningstiden i tankar med olika livsmedel.
Detta arbete gick ut på att använda olika metoder för att prediktera mixningstiden i en av Tetra Paks omblandade tankar, i syfte att förbättra deras förmåga att göra bra prediktioner. En av de använda metoderna var en empirisk metod där vi approximerade mixningstiden baserat på mixningsdata för standardiserade tankar framtaget på andra institutioner. Den andra metoden var mer sofistikerad, där vi utförde simuleringar av omblandningen genom det som kallas computational fluid dynamics (CFD) för att prediktera mixningstiden. Prediktionerna från dessa metoder jämfördes med experimentell data för att utvärdera metodernas prestanda.
Vi fann att en empirisk metod fungerade väl vid omblandning av lågviskösa vätskor (t.ex. vatten) i en relativt standardiserad tank, och att CFD inte var nödvändigt att använda i dessa fall eftersom det tar betydligt längre tid. För de högviskösa vätskorna vi testade (t.ex. yoghurt) lyckades ingen av metoderna förutsäga mixningstiden. Vi fann även att den empiriska metoden
fungerade sämre i mindre standardiserade omblandade tankar, där istället metoden med CFD gav bättre prediktioner och har mer potential.
Generellt kommer arbetet stärka Tetra Paks kunskap kring vilka metoder som är mest lämpliga att använda i olika situationer för att prediktera mixningstider när de designar deras tillverkningsprocesser. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9162472
- author
- Bågmark, Gustav LU
- supervisor
-
- Dragana Arlov LU
- Johan Revstedt LU
- organization
- course
- KLTM06 20241
- year
- 2024
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- CFD, mean age theory, stirred tanks, mixing time prediction, mixing time correlations, food engineering, nutrition and food chemistry
- language
- English
- id
- 9162472
- date added to LUP
- 2024-06-13 12:55:22
- date last changed
- 2024-06-13 12:55:22
@misc{9162472,
abstract = {{Stirred tanks are essential components in design and optimization of food processing lines, as they assure proper mixing and homogenization of ingredients. In this context, the mixing time is one of the most impactful performance parameters affecting various process design decisions. Thus, reliable methods to accurately predict mixing times in stirred tanks are highly desired.
The aim of this thesis is to expand Tetra Paks knowledge in mixing time prediction by using empirical methods and computational fluid dynamics (CFD) to predict the mixing time in one of their pilot tanks. Experimental mixing times had been produced prior the thesis, which are used to compare and validate the predictions against. CFD is used by developing a mixing time prediction method based on recent research that uses mean age theory (MAT).
CFD simulations of the pilot tank were conducted using two different agitators, one multi-impeller setup with baffles (MI setup) and one single-impeller setup without baffles (SI setup). The SM method was used in the simulations where various mean age distributions were extracted and used to obtain mixing time predictions, which were compared with the experimental data and empirical predictions.
For the MI setup, we found that an empirical method performs well in the turbulent and transitional regime. The CFD method also provides accurate predictions in the turbulent regime, and in the transitional regime with some method tweaks. This is however deemed less useful due to the accuracy of the less time consuming empirical method. Neither empirical methods nor CFD were able to predict the mixing time in the laminar regime. For the SI setup, we found that its dissatisfactory mixing performance makes the empirical predictions inadequate, and that the CFD method gave proper predictions with some method tweaks. Generally, we concluded that a CFD
method is valuable in the laminar regime and for the SI setup where empirical correlations lack applicability and performance. However, the tweaks required using the MAT method makes it significantly less robust, why other CFD methods are considered more suitable in comparison to MAT in its current state.}},
author = {{Bågmark, Gustav}},
language = {{eng}},
note = {{Student Paper}},
title = {{Mixing time prediction in stirred tanks using empirical methods and computational fluid dynamics}},
year = {{2024}},
}