This research evaluates the applicability of Multivariate Imputation by Chained Equations (MICE) for estimating missing well-log data across different sedimentary basis. Utilizing various machine learning techniques including XGBoost (XGB), Random Forest (RF), K-Nearest Neighbors (KNR), and Bayesian Ridge (BR), the performance of MICE was tested on three different data sets from distinct geological contexts and preprocessing conditions with minimal user input.
The main results indicate that the performance of MICE varied across different data sets and well-logs, highlighting the complexity of imputing missing data in heterogeneous sedimentary basins. The number of iterations in MICE did not significantly impact the performance of the models, while data quality, pre-processing, and geological complexities played crucial roles. The Force-200 data set, which underwent extensive preprocessing, demonstrated better imputation performance compared to the Montney and Beetaloo data sets. Additionally, XGB often outperformed other algorithms, predicting missing values with different number of iterations.
The main conclusions drawn from this study emphasize the need for more research to minimize user input and to develop more robust and flexible approaches to imputing missing data in well-logs. The study highlights the challenge of determining a single set of hyperparameters optimal for all the well-logs, suggesting the need for more adaptable models or even advanced techniques like deep learning techniques. The research also suggests the importance of refining pre-processing techniques, exploring further combinations of well-logs, and developing cross-validation approaches that effectively replicates real-world scenarios to advance the application and reliability of MICE in data imputation of subsurface data with missing values.

Measuring rock stress is a difficult process, especially in the Stjernøya Nepheline-Syenite mine in Northern Norway. The complex nature and topography of the shallow deposit makes it difficult to conduct industry standard tests. Therefore, this thesis aims at answering how different stress measurement techniques and their results compare to each other in a shallow low-stress hard-rock mine. To do this, an elaborate laboratory work is done to find a relation between stress (UCS) and acoustic properties and velocities of the nepheline syenite. Using that information, a similar combination of tests is done in the field, namely a flat jack test and acoustic velocity measurements with a hammer and geophones. To give further insight in the gathered field-data, visual classification methods, like RMR are done to verify and aid the tests and the results. The laboratory tests gave more insight in the relation between stress and acoustic velocity. There exists a positive, somewhat logarithmic relation between confining pressure and velocity through the sample, until the end of the elastic domain. Depending on the sample, this becomes more apparent after an initial loading phase between 0 and 10-20 MPa. In this low-stress zone, the wave arrival times could not be accurately observed and tend to give very low results. This problem translated to the field. Despite the flat-jack provided an understandable vertical pressure of 9.8 MPa, the acoustic measurements in the field did not always coincide with this pressure, according to the laboratory relationship, generally giving lower velocities than expected. However, a relation between rock quality and wave-velocity could also be present, as the cracks in the heterogeneous rock could heavily influence acoustic velocity. Therefore, this research concludes that both the flat jack and the acoustic investigative methods have their use in understanding the stress-profile of the shallow deposit. Given the low stresses in the mine, the acoustical investigative method could not be used directly to measure stress, but it can be used as a control for the flat jack tests and have use in monitoring of the rock wall.

In a world transitioning towards renewable energies and lithium-battery powered cars, there has been an annual increase in demand of 6% for lithium and other rare metals between 2000 and 2008 (Stringfellow & Dobson, 2021a), and this increase in demand is forecast to continue through 2040 (Latnussa et al., 2020). Meanwhile, there is a concern that supply will not be able to keep up with demand. Most of the lithium in the European markets comes from other countries such as Chile and Australia, making the European lithium supply vulnerable to supply chain disruptions, as was observed during the COVID-19 pandemic (Latnussa et al., 2020). In addition, the economic value of lithium has increased more than threefold compared in 2022 compared to 2017 (Latnussa et al., 2022).
The geological information of Northern-Western Europe at the latest stage, gives an explanation to the occurrence of lithium in the Netherlands.
50 water samples of 13 fields are considered for the chemical composition and origin of their brine. All geothermal brines contain dissolved lithium and other metals from the reservoir formation and surroundings, allowing for metal ‘mining’ in an environmentally friendly manner. The highest lithium concentrations in aquifer brines are in the province of Drenthe and Limburg, specifically in the Akkrum-13 gas field (47-48 ppm) and the Californie Geothermie (20-28 ppm). The correlation is positive only for lithium against rubidium and lithium against the vicinity of volcanic intrusions, from mineral mobilization through groundwater flow. There are no specific formations with high lithium concentrations.
In Germany, construction has started on a project to extract lithium from geothermal process water, indicating that it is economically viable (Wedin, 2022). For the Netherlands there are 5 different surface options to extract the highest lithium concentrations. Reverse osmosis, nanofiltration, ion-exchange, sorbents and electrodialysis can be used as direct lithium extraction methodologies. One potential economical method to extract lithium from geothermal brines is by utilizing two centralized lithium extraction plants. A plant for the direct lithium extraction with ion-exchange resins, placed before injection wells, and one for the refinement of the material.

The enlargement of the lock complex of Terneuzen in 2023 will allow Capesize vessels up to 300 meters in length, to access the Ghent-Terneuzen canal and reach the quay of ArcelorMittal Ghent (AMG). The goal of this thesis is to estimate to what extent these vessels can be used, the impact that these vessels will have on the supply chain and the cost impact of the implementation of Capesize vessels. The production of AMG is expected to be increased up to 5.5 million ton of steel for 2026. Therefore, it will require an estimated 12.9 million ton of raw materials (12% increase compared to 2019) which mainly consist of iron ore and coal. Up to 31% of this material can be shipped from source ports that are accessible by Capesize vessels. The draught restriction of 12.5 meters for the canal, limits the amount of bulk material that can be unloaded by the Capesize vessel to 105,000 DWT which is 60% of the maximum of 175,000 DWT. The remaining 70,000 tons of material can be unloaded externally and be transported to AMG by the use of inland barging vessels. The increased volume of Capesize vessels compared to the current Panamax vessels will lead to an increase in load size and a decrease in load frequency which will affect the downstream installations of AMG. Most of the current quay cranes will be unable to unload Capesize vessels due to the increased size of the ships. A proposed crane was introduced in the model to examine the ideal positioning to unload Capesize vessels and barges. 6 realistic scenarios for 2026 have been examined with varying levels of Capesize utilization. As a result of these scenarios vessel arrival lists for full year were simulated. Each of the vessel lists served as input data on a quay model with as results the downstream impact. The main output parameters were the waiting time for sea vessels, waiting time for barges and the amount of material that had to be stored on a temporary storage. In general, the implementation had a negative effect on all three of the output parameters resulting in additional quay costs. The net effect of Capesize vessels on the total costs was positive thanks to the lower shipping price per ton of material. This benefit was partly compensated by the increased costs for barging that was caused by Capesize transport. The shipping costs for Capesize vessels should at least be 2.53 dollar per ton cheaper, to break-even with Panamax vessels. Under current circumstances the full-scale use of Capesize vessels can be feasible. However due the high volatility of the dry bulk market and bunker prices, it is impossible to guarantee this will be the case for 2026.

The lime is a versatile product with multiple applications in everyday life. According to the European lime association, each citizen uses around 150 grams of lime per day. It is used as a base material to produce various products in environmental, agriculture, metallurgical production, construction, and chemical industries, among others. The high range of lime applications generates many limitations based on the content criteria set by the consumers and varies according to the requirement and further use of the lime. According to the European lime association (2017), one of the highest lime uses is on steel production; 40% of the total world production is sold to the steel industry each year. The quality requirements for the lime to be used to produce steel are strict, setting the target for the obtention of lime with a high concentration of CaO and low presence of prejudicial impurities such as phosphorus or sulfur. Calgov lime production plant is located in the south of Spain. It supplies local and national consumers of lime and has been operating for approximately 35 years. The steel manufacturing industry is one of the major clients of the plant consuming about 50% of the total installed capacity. This industry's quality requirements are a significant concern for the plant, and the obtention of limestone with low phosphorus content one of the primary targets for the company's future. Currently, the plant is under high pressure to find and extract limestone with low content of P2O5, below 0.039%, which is the limit set by the client. Seeking to comply with quality and quantity, the plant has performed exploratory work in the southern area of the deposit in which resources of limestone that meet this target are found. In this work, a mine design and scheduling options for the extraction of the south area of Calgov deposit extraction is presented based on the composition requirements set for the limestone. Furthermore, a techno-economic analysis of the extraction scenarios is developed exploring different alternatives to deal with the lack of information about some areas of the deposit and alternatives to ensure a constant feed of lime to the plant, avoid over costs and increase the possible economic return.

Over the last decade, interest in deep-sea mining of Polymetallic Nodules (PN) has reignited. It is driven by an increasing demand of metals that are critical for the digital economy and modern technologies such as batteries. Methods for resource estimation of PN are known and commercially available. There is however no standard method of a planning a deep-sea mining operation. This is an essential step to define mineable reserves. To address this problem, an existing concept has been applied in this thesis. This planning concept was inspired by agriculture and subdivides a mining license area into large scale mining sites, which are in turn subdivided into smaller, continuouslymineable areas calledmining fields. The subdivision is illustrated in figure 1. The contribution of this thesis is the development of a methodology for this concept. The concept was applied to a framework based upon a license area in the Clarion-Clipperton Zone (CCZ) in the Pacific Ocean. The mining will be done using a Nodule Collector (NC), which is a self-propelled machine that continuously gathers PN on the ocean floor and transports them upwards to a vessel at the ocean surface. Certain areas are deemed constrained by slopes or nodule types. Mining sites wherein the NC will operate were outlined within the license area using those constraints and Nodule Abundance (NA).Within the mining sites, a new metric was introduced for assessing the mineability for the NC: the Straight-Line Distance (SLD). The next step down from mining sites are the mining fields. The operation of the NC within them is approximated by two different driving patterns. Different field shapes are also analyzed, along with the impact of obstacles inside the field. Different methods of creating fields inside the mining sites were considered. The procedure relied upon the SLD to define the boundaries between fields. Two sets of field are created: with elongated shape and compact shape. The best set of fields was selected using the approximation of NC operation. Lastly, a method for determining the sequence of fields is developed that uses a Genetic Algorithm (GA) as optimization method. It takes into account constraints for mining rate, operational hours, nodule targets and travelling time. The methods outlined in this thesis are flexible and can be applied in a variety of circumstances. The selection of mineable areas turns out to be the must crucial step. The subsequent delineation of mining fields and sequencing does not make or break the profitability of the operation, although they are essential steps for the definition of mineable reserves.

The first size reduction stage in open pit hard rock mining is blasting and is fundamental for mineral resource extraction as it enables transportation of the rock. Consistent frag-mentation results are preferable as it can ease the loading, hauling and crushing stages. Fragmentation analysis assists in identifying areas where similar blast results appear when comparable drill and blast designs are used. Subsequently, drill and blast domains can be defined for the Leveäniemi mine to ease the drill and blast design process.

Measuring fragmentation has been continuously researched over the past years. Image analysis methods were developed as it minimised disruption to production and provided a reasonable indirect estimation of particle sizes. So far, research was focussed on 2D image analysis. However, by adding a 3rd dimension, some limitations can be overcome that were experienced when using only two dimensions. Therefore, the potential of 3D image analysis of blasted rock, loaded in haul trucks, is of interest. By evaluating truck loads, the muck pile is better represented in comparison to measuring a whole muck pile. This is one of the few studies done until now on 3D image analysis in open pit mining, analysing material in loaded trucks.

A 3D image analysis field test was executed to measure blast-induced fragmentation in a production environment at the Leveäniemi mine. The aim was to gain a clear understand-ing of the factors contributing to an optimal blast result and to establish blast domains. The test setup consisted of an image acquisition system, photographing truck loads from above using two cameras that were triggered by a laser. RFID truck markers were com-bined with Minestar data to identify the origin of the truck. Truck loads were analysed using software developed by LKAB and 3GSM, constructing 3D models and automatically delineating particles to analyse fragmentation. No pre- or post-processing of the images or delineation results have been done. Drill and blast, and muck pile shape parameters were acquired as well.

The results show x50 particle sizes ranging from 5 to 56cm and x80 ranging from 20 to 150cm. Care should be taken when interpreting these results due to the limited amount of data analysed and bias in the measurements and software. Fines are underestimated and correct particle delineation occurred on average in 42% of each load. The limited amount of data resulted from the practical problems arising during data acquisition. Con-tinuous data acquisition of images was not achieved.

Correlation of fragmentation to drill and blast, or muck pile shape parameters was not achieved due to the challenges faced. 3D image analysis of truck loads proved to have potential but requires many modifications and developments to the system and software to achieve continuous data acquisition. A focus on image quality and the practicalities of the system is recommended. Continuous data acquisition is required when using frag-mentation analysis for establishing blast domains.

Comparable and repetitive measurements are the main prerequisites for choosing a method to analyse fragmentation for the purpose of defining blast domains. Hence, other methods like using drones to analyse whole muck piles after blasting should be consid-ered, though it would statistically be less representative. For the Leveäniemi mine, it is recommended to achieve geotechnical domains including joint spacing and orientation before a follow-up study is started on blast domains. Additionally, the development of this method should be finalised, or an alternative is recommended.

The transition from open-pit to underground mining operations is a frequently occurring procedure for mines around the world. The Akyem Gold Mine in Ghana is an example, where the transition is to be optimized with regard to the significant reduction in yearly throughput. The current processing plant is commissioned to efficiently process up to roughly 7.7 Mt of ore per year. The planned underground mine is likely to start operations in 2022 during the
final stages of the remaining open-pit operations, with an estimated yearly production of 1.6 Mt of ore. This implies that a major plant modification is recommended in order to prevent unnecessary high operating costs, once a full transition may occur. However, opportunities are seen to discard the plant modification and instead to add incremental open-pit material, which is currently considered as mineralized waste. The suggested option will be considered once the analyses prove that a certain incremental open-pit head grade will generate more value in combination with the underground activities. This analysis is based on financial modelling of the potential options, where a major trade-off is conducted between the plant modification for the planned option, and the incremental open-pit mining, processing and tailings storage for the suggested option. The analysis is tested by changing all different decision parameters to indicate the optimum solution. Prior to this head grade determination, blending optimization of the potential sources of ore is conducted, in order to determine the optimum period where the potential plant modification may occur. Results show that the open-pit mine will directly supply the processing plant until 2023, whereas its stockpiles will be depleted by the end of 2025. For this reason, the suggested plant modification is suggested to take place at the start of 2026. Subsequent results related to the option evaluations are in favor of the full transition to only underground mining. Despite any changes in the set of decision parameters and gold prices, the incremental open-pit head grade will not be possible to obtain. This results from the fact that the calculations indicate a deficiency of required gold ounces based on the remaining open-pit reserves. It may be concluded that the potential transition from open-pit to underground operations with a reduction in yearly throughput, is optimized through a full transition to underground mining. The planned option, with the cohering plant modification, is proven to be optimum and is currently considered as the solution for future mining at Akyem Gold Mine.

An investigation in the possibility to integrate fleet selection (through FPC) in the mine design process, allowing for the analysis of various potential haul road scenarios and from there being able to make a decision on an optimal fleet selection.