Aaltodoc

Aaltodoc is the institutional repository of Aalto University.

Aaltodoc has a new updated appearance. Instructions for searching and more information is available here.

Otakaari 1 grandhall. Photo: Esa Kapila
 

Communities in DSpace

Select a community to browse its collections.

Now showing 1 - 13 of 13

Recent Submissions

Data driven approach of understanding the relationship between vibration and wear rate to low-consistency refining process variables
(2025-04-16) Heikkilä, Henri
 School of Chemical Engineering | Master's thesis
The goal of this thesis is to evaluate how well data gathered with a continuous IIoT based monitoring system can be applied to monitor refiner plate lifetime and find out if the measured variables can be used to indicate changes in process conditions or refiner performance. For example, in refiner maintenance and operation optimization. Seven different real industrial refining cases are evaluated. From each vibration and plate position data are measured continuously. These were then compared against other mill process variables, measured in trial runs, to understand how these correlate with other process variables and refiner performance.Results showed a correlation of over 0.68 between wear rate and vibration with gross power, feed consistency SRE and each other. This indicates that vibration and plate position can be used as an indicator for changes in the refiner system. Between different refining cases the plate pattern and used furnish effected the wear rate and vibration the most. To further utilize the gathered data, methods used for forecasting the plate wear such as linear regression, ARIMA, and exponential smoothing techniques were evaluated. In most cases the wear was mostly linear, and a simple linear regression was enough to get a good forecast. Adding more variables or using more complex models increases the model accuracy, but the forecasts compared to actual data only had marginal improvements. ARIMA and exponential smoothing methods required data from a longer period of time to produce a reasonable forecast. The ARIMA worked reasonably well, especially with uneven wear rates. Future work should focus on testing these models with larger datasets to increase accuracy and further validating them using data over multiple plate lifetimes for the same refiner as well as more measurements in different types of refining cases. For example, determining the effect of consistency to vibration while using bleached kraft pulp as furnish.
Enterprise architecture as consultative tool in the strategy implementation
(2025-04-21) Hämeenoja, Henrik
 School of Science | Master's thesis
Organizations must align strategic objectives with operational capabilities in today's dynamic business environment to maintain agility and competitiveness. This study presents the Agile Enterprise Architecture (AEA) framework, an agile consulting model focused on addressing strategic IT and digital business challenges within Finnish consultancy practices. The AEA framework integrates agile methodologies with enterprise architecture (EA) to overcome traditional EA's rigidity, offering a structured yet flexible approach to strategy implementation. Through structured client engagements lasting three weeks to three months, the framework applies sprint-based consulting to rapidly produce targeted EA artifacts that resolve strategic IT challenges. It aligns existing methodologies like SAFe, Scrum and TOGAF while providing a more adaptable alternative for way of solving specific consulting tasks. The engagement process includes preparation, planning, sprint execution, and review, ensuring continuous improvement through iterative feedback. The research followed the Design Science Research methodology and incorporated qualitative semi-structured interviews, with the resulting data analyzed using the Gioia method (Hevner, et al., 2004; Gioia, et al., 2012). The resulting artifact produced was assessed and refined using the Illustrative scenario method (Peffers, et al., 2012). The study's reliability and validity were ensured by following these systematic research methodologies. By embedding agility into EA practices, the AEA framework facilitates quicker decision-making, reduces unnecessary documentation, and enables effective strategy execution. The framework's structured approach helps translate high-level strategy into actionable outcomes, reinforcing EA's role as a strategic consulting tool.
The reactions of carbon dioxide in alkaline media – Lignin recovery from black liquor
(2025-04-28) Raulahti, Anni
 School of Chemical Engineering | Master's thesis
This thesis investigates the use of carbon dioxide (CO₂) as an acidification agent to precipitate lignin from black liquor, a highly alkaline byproduct of the kraft pulp-ing process. The study aims to understand the behaviour of CO₂ in alkaline envi-ronments and improve the efficiency and control of lignin recovery by adjusting the process parameters. Particular focus is on the relationship between operational parameters and lowering the pH, which is critical for initiating lignin precipitation. A series of experiments were conducted to examine how process variables such as CO₂ flow, black liquor feed rate, reactor pressure, and feed temperature affect the pH lowering rate, CO₂ consumption and chemical composition of the product. The results demonstrated that all of these parameters influence the precipitation process, but CO₂ flow rate and black liquor feed rate had major impact on pH. Increased pressure and temperature generally accelerated the pH decrease. However, medium pressure had the overall best results. Notably, the reactions in black liquor differed from those in pure NaOH systems likely due to the presence of organic matter and other ionic species, which can affect both solubility and re-action pathways. Thus, research using black liquor is essential. It is concluded that CO₂ is a viable and efficient reagent for lignin precipitation, but continuing the optimization offers further benefits. The findings contribute to a bet-ter understanding of CO₂ reactions in complex alkaline media and provide data for improving the lignin recovery processes in industrial applications. Recommenda-tions for future work include filterability, improved gas-liquid dispersion methods and comparative studies using simplified alkaline systems to isolate key reaction mechanisms.
Typical polyolefin infrastructure utilization for mechanically recycled polymers production
(2025-04-28) Planting, Arthur
 School of Chemical Engineering | Master's thesis
Mechanical recycling is currently the preferred recycling method for plastic waste, due to lower operating costs than chemical and advanced recycling methods. In mechanical recycling the plastic waste is ground into recyclate and then extruded either as is or mixed with a virgin polymer. Chemical and advanced recycling is seen as complementary methods and relies on thermal processes, such as pyrolysis or gasification or other chemical processes, such as hydrogenolysis or oxidation to break down plastic waste into monomers or hydrocarbons. The high operating costs associated with chemical and advanced recycling are due to high operating temperatures, complex reactor design and fast catalyst deactivation. The economic potential is not a driving factor for recycling plastic waste, instead the driving factor stands from extended producer responsibility (EPR) programs, which are govern- mental initiatives or laws stating that the plastic manufacturer must recycle or treat the generate waste. Different process alternatives for producing polymers from mechanically re- cycled polymers were identified. The polymer is produced from either polymer A or pellets, which is transported to the compounding plant where it is mixed with recy- clate and compounded. Different implementation scenarios were studied and based on infrastructure utilization, complexity, investment profitability and environmental health and safety aspects the most viable option for respective process was de- termined. The most viable option for the polymer A process is to construct a new silo at the compounding plant, as the polymer A requires a special atmosphere, which limits the infrastructure utilization level. The most viable option for the pellet process is to utilise a silo at the LDPE plants silo farm to store the pellets before the production run starts. These were compared with a base case, batch transportation of pellets as this scenario does not require any capital investments. Although the polymer A case would require a significantly higher capital investment, it is the most viable option due to lower operating costs, increased product properties, higher profitability potential, lower safety risk, reduced environmental impact and increased capacity for the polypropylene plant and the compounding plant. This results in increased potential for expanding Borealis mechanically recycled product portfolio, BorcycleTM M, which would positively contribute to company growth.
AI and Value Stream Mapping
(2025-04-24) Poteri, Juho
 School of Science | Master's thesis
This thesis studies the connection of Value Stream Mapping (VSM) and AI, especially generative AI (GenAI), from a dual perspective. Firstly, considering that there is a growing interest towards increasing the efficiency of work with AI, but simple and readily available AI tools that target the individual can only provide easily imitable benefits, it is evident that there should be a more structured approach as to how AI could be integrated to business processes on a deeper level. To this end, the thesis discusses how VSM could be applied to provide structure into the effort of integrating AI into business processes. Secondly, as these process mapping and improvement tasks can be rather laborious, a small AI agent system is developed and tested as a support for this VSM work. AI in this work is seen as black box type of tool that is characterized by its ability to process, and in the case of GenAI also output, unstructured data, but in a seemingly non-deterministic manner. The three ways in which it can be used in human tasks, according to the literature are: full AI automation, sequential combination, and AI-human hybrid combination. The characteristics of these, in terms of for example the depth and innovativeness of the results produced differ. To consider what kind of integrations would be suitable in different places within processes, the usual metrics such as process and lead time, are accompanied with two new metrics, repetitiveness and significance that roughly describe how suitable a step is for automation and what the associated risks are. High repetitiveness would support full AI automation, except if significance is high as well, in which case a sequential combination would be preferable. For lower repetitiveness the hybrid combination would be suggested. The agent system developed in the work that supports the mapping and analysis of value streams is composed of three agents that focus on different stages of the analysis, and a graph-based RAG implementation that allows the agents to digest information from documents that describe the processes to be analysed. The agent system demonstrates remarkable capabilities in visualising and analysing processes, but some aspects, especially the accuracy of computations, still seem prone to errors in highly complex situations. Overall, the VSM framework extension and the agent system exhibit potential, and especially the integration of process mining capabilities to the agent system would further enhance its capabilities.