���ʵ���

Author: Daffa Reza Kaiyandra 

Supervisor: Mikael Rinne

Advisor(s): Masoud Torkan, Mateusz Janiszewski

Funding: - 

Abstract: 

This thesis tests whether AI-based methods applied to 3D point clouds can reliably support rock mass classification. A 3D point clouds of a 2.6 m × 4.6 m drift wall that was produced using photogrammetry was processed in CloudCompare and Python.

Point normals were clustered using unsupervised clustering method in Python. KMeans and Mean-Shift is implemented on the unit normal vectors to identify joint sets and their mean orientations. Within each set, DBSCAN was applied to group spatially continuous patches, and RANSAC was used to extract representative planes for spacing analysis.

Small planar facets, generated with the FACETS plugin in CloudCompare, were used to estimate surface roughness via JRC₂₀. Joint spacing and volumetric joint count were obtained from the RANSAC planes and a custom tool in Python that measures perpendicular distances between planes along the mean set normal.

These results were compared with traditional scanline and compass measurements collected on the same wall. The semi-automatic workflow reproduced the three main joint sets within a few degrees of the manual orientations, gave RQD values close to the measured 77.5%, and produced Jr values in line with field roughness profiles. When used as inputs to Q, RMR, GSI and RMi, the parameters that are obtained from point clouds led to rock mass classifications in the same “good rock” range as the manual assessment.

Overall, the work shows that AI-based methods can provide as consistent input data for rock mass classification, serving as a valuable complement to conventional field mapping.

Keywords: photogrammetry, 3D point clouds, URLA, artificial intelligence, machine learning, clustering, rock mass characteristics, rock mass classifications

Author: Manish Jaiswal

Supervisor:  Prof. Jouni Punkki

Advisor(s): Senior Advisor Leena Korkiala-Tanttu , DSc. Anoosheh Iravanian 

Collaborative partner City of Helsinki

Abstract: 
This thesis investigates the use of recycled materials, such as incineration slag and fly ash, in rammed earth construction as a sustainable alternative. The study focuses on evaluating the strength, durability, and moisture resistance of stabilized rammed earth mixes, with incineration slag as the aggregate and fly ash as the binder.

Various binder-to-aggregate ratios, moisture content, and aggregate mixes were tested. Key tests included uniaxial compressive strength (UCS), freeze-thaw resistance, ultrasonic pulse velocity (UPV), and calorimetry. The best performance was observed in a mix with a 1:4 binder-to-aggregate ratio and 16 % water, using incineration slag, crushed concrete, and Nilsiä sand, achieving a UCS of 11.1 MPa after 28 days. In comparison, mixes with only incineration slag reached 5.21 MPa. After 15 freeze-thaw cycles, the UCS of the 1:4 incineration slag only mixes increased by 41 % to 7.33 MPa, while the 1:5 mix increased by 33 % to 5.96 MPa. Hydrophobization reduced capillary water uptake by about 96 %.

Calorimetry tests indicated that higher binder content and adequate water significantly improved hydration. The 1:4 mix with 20 % water released about 220 J/g binder after 72 hours, while the same mix with 17 % water released 190 J/g binder, indicating more complete hydration and better pozzolanic activation of the fly ash. Further research is needed to assess the long-term performance of hydrophobization treatments and moisture control, as well as their application on a larger scale.

Työn nimi:       Last Planner –menetelmän tehostaminen suunnittelunohjauksessa           

Tekijä / Author: Jussi Viemerö

Valvoja / Supervisor: Jussi Leveinen

Ohjaaja(t)/Advisor(s): Sami Rämänen

Rahoitus/Funding: ei/no

�վ����������ٱ����ä/�������ٰ�������:  

The main challenges in the construction industry are seen as planning and in particular design management. Studies show that a significant amount of waste is generated during design phase, which becomes evident in the construction phase as incomplete, incon-sistent, and un-constructible plans. However, design management methods adopted in the industry have demonstrated that, when properly implemented, they can significantly improve the overall success of construction projects.
The objective of this thesis was to determine how the LPS design management meeting can be made more efficient. The effectiveness of an LPS meeting can be measured using the Percent Plan Complete, known as the PPC value. The study examined the current state of the case company’s LPS meetings by calculating the PPC values of both completed and ongoing projects and by identifying the types and causes of waste occurring during LPS meetings.
The research was conducted by utilizing mixed method approach which combines both quantitative and qualitative research approaches. Using the mixed method enabled the phenomenon to be examined through both objectively measurable data and subjective observations and experiences. The quantitative part consisted of design tasks recorded in LPS meetings across 12 case projects, while the qualitative part included observations of LPS meetings in five projects and five themed interviews.
Based on the quantitative findings, no clear or consistent relationships were identified between PPC values and the categories of incomplete tasks. The observed correlations were weak and partly random. These results suggest that variations in PPC values are more likely explained by the combined influence of multiple factors rather than by any single category of incomplete tasks.
The qualitative analysis identified 17 forms of waste in LPS meetings and 25 underlying causes. The most significant issues were related to the lack of standardisation in the LPS operating model, ambiguity of tasks, insufficient meeting time, inadequate decomposi-tion of design tasks, and challenges in obtaining reliable promises for tasks and recognis-ing task dependencies. The study presents several measures for reducing or eliminating these sources of waste, thereby improving the efficiency of the LPS process. Additionally, the research produced an Excel database compiling the design tasks of all 12 case projects, which remains in use within the case company.

Author: Matti Hopia

Supervisor: Mikael Rinne

Advisor(s): Vesa-Matti Matikainen

Funding: 

Abstract:

The aim of this thesis was to determine what kind of information about the bedrock is necessary to obtain to evaluate a drilled steel pipe pile’s geotechnical strength, and how that information can be obtained in a way that’s both practical and reliable. Another goal was to identify the factors that might increase the risk of pile settlement when the structure is under load. As ways to obtain information about the bedrock this study looked at dynamic loading tests of piles, observations made during the pile’s drilling process and different geological survey methods. Data from different projects were analyzed each method. The study also included a literature revie w and interview with a pile driller to support the findings. Results of this study indicated that poor rock mass quality (Q’=0…1) at the pile tip increases risk of settlement. Based on this, it is important to make sure that the rock mass quality is good enough at the pile tip and up to a certain depth under the pile. The most reliable method of determining the geotechnical strength of a pile is the dynamic loading test. It is challenging to name the most practical one based on this study. The examined methods support each other, which is why quality assurance should continue to be a combination of these methods. The used combination of methods should be evaluated based on the project’s needs.

Author: Hang Liu

Supervisor: Prof. Mikael Rinne

Advisor(s)/Co-supervisor(s):Dr Mateusz Janiszewski and  Tuomo Hänninen

Funding: No

Abstract: 

The thesis investigated the development and evaluation of a virtual blasting training (VBT) system for underground tunnels. VBT is based on 360-degree photos of a tunnel blasting simulator in the Underground Research Laboratory of Aalto University (URLA) and blasting course materials. The goal was to improve blasting knowledge and skills in a safe, digital environment. The system adopts a dual platform architecture mode and utilises ThingLink and 3DVista Virtual Tour Pro software to develop six blasting scenarios. ThingLink was used for safety induction, while 3DVista Virtual Tour Pro was used for more complex blasting program training. For example, blasthole type and placement, charging and fuse connection, blast initiation, etc. The system was evaluated through feedback and questionnaires from a blasting expert and 10 users. The results showed an improvement in safety awareness and blasting knowledge. Thus, VR technology provides a new possibility for underground tunnel blasting training and is expected to be applied in future practical training.

Author: Puja Rajthala

Supervisor: Associate Professor Wojciech Sołowski

Advisor(s): Msc. (Tech) Jani Lepistö, Msc. (Tech) Erkki Liimatainen

Funding: Insinööritoimisto Lepistö Oy

Abstract

Settlement estimation in layered fine-grained soil is vital for choosing an appropriate foundation type for the infrastructure structures, including buildings and highways. This thesis focused on predicting the time-dependent primary consolidation behaviour of layered fine-grained soil in the Kujala area, where settlement data have been monitored over the past two years. To acquire this, Janbu parameters- including the modulus number (m), stress exponent (β), and preconsolidation stress (σ_p) were determined by curve-fitting the stress-strain curve obtained from oedometer laboratory tests conducted on undisturbed soil samples from the Kujala Interchange area. Initially, the coefficient of consolidation (C_v) for each soil layer was acquired using Taylor’s curve-fitting technique for the average degree of consolidation (U) versus time factor (T_v). Furthermore, to account for uncertainties resulting from inherent variability and sampling processes, characteristic values of Janbu parameters were evaluated for each soil layer, and the value of C_v was altered until the monitored time-dependent primary consolidations were replicated. 

The results indicated that the monitored time-dependent consolidation behaviour of the study area could be successfully replicated by varying the coefficient of consolidation (C_v) using the Tangent Modulus method. However, there is a significant deviation in C_v value derived from the laboratory test from that occurring in the field since the drainage boundary condition, anisotropic permeability, and stress history of soil might not be the same as in the field while conducting the laboratory test. Hence, Janbu’s settlement model might be adopted with caution and is reliable only if the monitored settlement data are available to support back calculations.