The case study of the outer lock head of the New Terneuzen Lock is considered to investigate the behavior of the Boom Clay and to develop an automated method for improving the deformation predictions in a deep excavation format. The initiative to perform this study was the uncertainty in modeling the Boom Clay behavior during the design and the over-predictions of the combi wall deformations in the aftermath. The problem has been attributed to the constitutive soil model and soil properties used for that layer. In the context of this thesis a Python - PLAXIS application has been developed, which facilitates an iterative method to determine the real soil properties of the Boom Clay layer based on inclinometer measurements of the wall throughout the construction stages. The monitoring data present the reality; hence the safety factors had to be filtered out of the design for the comparison with them to be applicable. Therefore, the mean (most probable) soil properties have been recalculated, and the as-build external loads, aquifer heads, structural elements, and phasing have been deciphered to create the Mean model. Using the Mean model, the most appropriate constitutive model and drainage conditions for the Boom Clay are determined. It has been concluded that the Hardening Soil small strain is the most appropriate constitutive model. Using the Mean model as initial input and the constitutive model conclusions, an iterative process on the relevant soil properties has been conducted to reach ±10% convergence with the corresponding monitoring deformations. The Fine Tuned model uses the soil properties derived by the iterative method to represent reality with the highest accuracy and reaches, on average, 45 % more precision on the prediction of the deformation in comparison to the monitoring data than the Mean model. Using the Fine Tuned model, sensitivity analysis of the wall to Boom Clay’s soil properties is performed. It revealed that φ′ had the highest impact relative to the other properties. Scenarios with thinner Boom Clay layers have been tested due to the relevant wedging geometry it follows in the project location. Additionally, the Fine Tuned model has been used to improve the prediction of future stages with data derived from earlier monitoring. The example studied was able to improve the prediction by 87 % in comparison to the initial design. In comparing the Design model with the Fine Tuned model, a problem with the anchor wall behavior is discovered that is attributed to the inability of PLAXIS to consider the shaft friction of the anchor rods. The Fine Tuned model, in combination with the fixity solution, allowed for an improvement in deformation accuracy of up to 95 %. It is concluded that the actual design, despite over-predicting its deformations, produced a retaining wall validated by the Fine Tuned model. It is suggested that the Python Application should be used alongside the traditional designing process and in site engineering to reduce the risk by simulating the actual behavior and limits of the retaining wall.