With the implementation of options to couple programming languages and geotechnical finite element method software, the potential for optimized design in geotechnical engineering has arguably never been greater. However, knowing the situations in which these possibilities are the most useful is not always clear. This thesis has attempted to answer this by using Python in PLAXIS 2D to model and cost optimize single anchor sheet piles at the new Drammen Hospital in Drammen, Norway. Along with the analysis itself, the general viability of modelling and analysis using Python in PLAXIS 2D for regular geotechnical engineering practice has been evaluated.
The study starts by looking at previous optimization studies within the field of geotechnical engineering, evaluating the different methods with regards to their benefits and drawbacks. Some general optimization theory has also been presented. For a comparative analysis, three optimization methods were selected: an optimization algorithm from a Python library, a simplistic optimization script and an automatized brute force method. The modelling and analysis of the profiles in the thesis has been done purely using Python scripting, in order to give a good idea of its potential in regular engineering practice. A focus throughout the study has been to look at where and when the use of scripting is the most effective.
The results from the study show that there is good potential for cost optimization using Python in PLAXIS 2D if the different construction costs are available to the engineer. The results indicate that finding a solution in the transition between the two most dominant failure mechanisms yields the most cost-efficient solutions.
Modelling using Python in PLAXIS 2D is found to have limited viability for regular practice in geotechnical engineering, as many of the actions, such as drawing irregular geometry, changing boundary and water conditions and creating materials are all easier done manually. Materials especially should always be created manually, as the available manual tools are doing it through scripting can be tedious and is prone to error messages. Overall, scripted modelling is found to have more drawbacks than possible advantages. That said, for more research-based projects, automatized modelling could still be highly useful.
Analysis using Python in PLAXIS 2D is found to have a lot of potential and great prospects. While there are a few drawbacks that must be accounted for, the possibilities offered by automatized analysis are found to outweigh these. Being able to run highly detailed analyses overnight is in itself thought to be very useful in terms of efficient analysis. Combining this with robust optimization scripts can allow for overall better designs and lower design costs. The benefits of cost-optimized design will increase as project size increases, making it particularly useful for large projects. In the case of small projects, the extra time spent optimizing design must be weighted against the potential gain.