A civil engineering work can be performed by organizing theavailable resources (manpower, equipment and materials) inmany different ways. Each different configuration results in arealization time and a cost that a building company has to bear.To produce reliable duration forecasts and money savings, itis essential to take into account all the uncertainties involvedin the project operations. Generally, since it is impractical toprocess numerous uncertain variables - also undefined froma statistical point of view -, traditional probabilistic methodsinvolve application difficulties for complex environmentssuch as construction sites. To properly handle this issue, theauthors propose in this paper the application of the AffineArithmetic technique. This method treats the variables as intervalsand returns reliable results, even when the variables aremutually dependent. The numerical example presented in thepaper proves the efficiency of the procedure, even if some analyticalcomplications are included in the analysis (dependencybetween variables, non-linear functions, etc.). Comparisonswith Interval Analysis and traditional procedures are also provided.Adopting Affine Arithmetic, the results are reported interms of intervals, avoiding the definition of unrealistic deterministicvalues that can strongly affect the operation organization.Furthermore, without increasing the problem complexity,the model admits continuous modifications (interval amplitudes,new variable dependencies, etc.) to correct and optimizethe durations.