Based on previous studies, the Dutch Building Regulation (BBL) is inadequately equipped to address fire safety in timber-structured buildings, particularly when the timber is exposed and in the context of tall buildings. Fire compartments containing large volumes of structural timber suffer from additional fire safety risks compared to traditional buildings without a combustible structure. These extra risks include: a longer fire duration, unpredictable fire scenario, potential second flash-over, failure of timber members due to heat wave-penetration, larger protruding flames from openings and increased smoke production (Brandon et al., 2022). This master thesis focuses on studying fire safety of CLT (Cross Laminated Timber) compartments with properties typical for apartment buildings. The master thesis contains a literature study, divided into material level, structural component level, compartment level and building level. This literature study can be helpful for people who want to understand the fundamentals of fire safety in contemporary CLT timber buildings and want to understand the further master thesis with more ease.
In this master thesis we used three fire behaviour models to analyse compartment fires previously tested in full scale by Brandon et al. (2021a). The outcomes of simulations done with these models were compared to full-scale compartment fire test data (for a compartment of 48 square meters) to get an understanding of the accuracy, and strengths and weaknesses of the models. The study suggests that the ZHM and Brandon models are able to simulate the average char depth in the ceiling. For compartments that have percentage of exposed mass timber and an opening factor equal to that in test 1 of Brandon et al. (2021a), the outcomes of the Brandon and the ZHM models give conservative results for the the maximum char depth in the ceiling. There is a discrepancy between the simulation results and the maximum char depth in the walls when amount of exposed timber and opening factor were increased beyond the values used in test 1 of Brandon et al. (2021a). Suggestions for correction factors based on linear interpolation are given to arrive at a better approximation for the maximum char depth. It is advised to use CFD and/or compartment fire tests for any compartments that differ from tests 1-5 in Brandon et al. (2021a). Targeted design measures were listed to protect the critical areas where maximum char depth is expected.
In order to get a better understanding of the fire behavior models for CLT compartments, I simulated 24 different realistic compartments, with differing sizes, shapes, amounts of exposed timber and opening factors. The parameters were chosen in such a way, that the variations encompass a range of typical CLT apartment buildings. I performed this analysis using three different fire behaviour models. The models calculate the charring depth into the CLT cross-section. We then added the zero strength layer (timber without any structural resistance due to the heat-wave) to the modelled charring depth, resulting in the effective char depth. The larger the effective char depth, the more of the cross-section’s initial resistance is lost. Over the three models, the largest correlation was found to be between opening factor and effective depth. The second largest correlation was found to be between amount of exposed timber and effective depth. Using the results of the models, without any correction for maximum char depth, none of the simulated compartments suffer failure of the ceiling panel according to calculation.
Between the three models, the zone model (Brandon et al., 2021b), is best capable of simulating protection for a finite number of minutes, as the exact number and thickness of the protection can be input in the model.

Concrete compressive strength is the most frequently used and most important mechanical propertyof concrete. National and international building codes (such as the Eurocode) frequently use the com­pressive strength for design with concrete.

In some cases, instead of testing the concrete specimens under compressive loading in laboratories, predicting the compressive strength using machine learning predictions could be a good alternative. The ingredients making up the concrete mix and the curing age can be used as predictors for the com­pressive strenght.

Artificial Neural Networks (ANNs) are machine learning algorithms that have been used since the nine­teen sixties and were inspired by the way neurons work in the brain. Previous researches suggest that ANNs have great potential to predict concrete compressive strength.

In this research, an Artificial Neural Network was set up using the Keras framework and was trained with concrete composition data consisting of examples of concrete recipes and their respective con­crete compressive strenghts. Then, three hyperparameter optimization methods (for loop, grid search and Bayesian optimization) were implemented in several runs. The resulting hyperparameters were used to create ANNs. Afterwards, the three methods were compared with respect to several met­ rics (R­squared score, root mean square error and running time) to see which one is the relatively best method to provide hyperparameters for the predefined ANN that learns from concrete composition data.

The best runs of the three hyperparameter optimization methods show similar goodness­of­fit (with negligible difference). Among these best runs, grid search has the shortest running time. Bayesian optimization provides the highest R­squared score, and has root mean square error of 4.45 [MPa] and a reasonable running time of 73 minutes. Therefore, Bayesian optimization is considered the preferable hyperparameter optimization algorithm for the concrete compressive data used as goodness­of­fit is in most cases to be considered the decisive metric for this application. Further research trying runs with larger parameter grids and spaces, or using other tuning methods could result in even better goodness­ of­fit results.