Prediction of Repair Work Duration for Gas Transport Systems Based on Small Data Samples
Volume 12, Number 4, July 2016 - Paper 1 - pp. 305-320
VALERY LESNYKH1,2, YURI LITVIN1,2 and IGOR KOZINE31Science Research Institute of Economics and Management in Gas Industry, Moscow, Russia2National Research University “High School of Economics”, Energy Institute, Moscow, Russia3Technical University of Denmark, Department of Management Engineering, Copenhagen, Denmark
(Received on June 07, 2016, revised on June 25, 2016)
Prediction of the duration of a repair and maintenance project of a gas transport system is an important part of planning activities. There exist numerous sources of uncertainties that may result in time overruns possibly leading to multiple negative consequences. Our experience in planning this work suggests that accepting the stochastic nature of the project duration is a constructive step towards the preparedness to contingencies and defining penalties for repair companies. To support this approach, one needs to construct probability distributions of the durations of the projects. To address the issue of the scarcity of observed data, we suggest using a bootstrap resampling procedure. Gram-Charlier functions and order statistics are employed to approximate the distributions. It is demonstrated how to derive them for a separate repair project and a larger project consisting of a number of concurrently running subprojects. Following this, guidance is provided on how to decide about what duration should define the deadline for completion of the whole work. A simple example is provided.
Click here to download the paper.
Please note : You will need Adobe Acrobat viewer to view the full articles.