Non-linear optimisation in planning and scheduling of earthwork projects

Abstract

Construction of new motor and railways involves hauling of millions of cubic metres of soil along the project road line. The total cost of earthworks often reaches several million Euro leading to a strong need for project optimisation. Small gains in mark-up are welcomed both because they are worth hundreds of thousands Euro and because the earthwork market is very competitive.

In order to execute the work, companies have at their disposal numerous solutions for each problem. Materials can be moved with different resources. Bad materials can be replaced, treated, drained in different ways. Each solution will be interesting depending on constraints but each solution impacts on other tasks. If a soil replacement is decided, bringing good material will be needed and it cannot be used elsewhere and a disposal area has to be found. On the other hand, a soil treatment may affect hauls on the section. Thus projects can only be optimised as a whole.

This study compares optimisations currently done in France and in Finland. More specifically, it focuses on optimisation of mass haul plans. Mean distance minimisation - a linear optimisation - has been used for more than twenty years in France. Non-linear cost optimisation has recently been introduced in Finland. Optimisation algorithms themselves are not studied but the methods and data needed in the execution of the different optimisations. Important parameters are reviewed as well as how to take advantage of non-linear optimisation with respect to the French earthwork project specifics compared to current state of linear optimisation schemes.

This study has been carried out with technology acceptance in mind, as measured by the technology acceptance model (TAM). Thus simple examples have been built to improve perceived usefulness.