Assesing the efficiency of reducing variance methods in construction project network simulation
Fragment książki (Rozdział monografii pokonferencyjnej)
MNiSW
5
spoza wykazu
| Status: | |
| Autorzy: | Biruk Sławomir, Jaśkowski Piotr |
| Dyscypliny: | |
| Aby zobaczyć szczegóły należy się zalogować. | |
| Wersja dokumentu: | Drukowana |
| Arkusze wydawnicze: | 0,5 |
| Język: | angielski |
| Strony: | 45 - 50 |
| Efekt badań statutowych | TAK |
| Materiał konferencyjny: | TAK |
| Nazwa konferencji: | 8th International Conference from Scientific Computing to Computational Engineering |
| Skrócona nazwa konferencji: | IC-SCCE 2018 |
| URL serii konferencji: | LINK |
| Termin konferencji: | 4 lipca 2018 do 7 lipca 2018 |
| Miasto konferencji: | Ateny |
| Państwo konferencji: | GRECJA |
| Publikacja OA: | NIE |
| Abstrakty: | angielski |
| . Monte Carlo simulation is a popular tool that supports planning projects affected by risk (among others, for defining project completion time at certain levels of probability, or assessing the impact of activity modes to select best options). Computer simulations enable the planner to formulate and verify hypotheses on distribution type and parameters of scheduled event occurrence and the project's duration. Monte Carlo simulation testing of network models, which feature a deterministic structure and comprise activities the duration of which are random variables, do not require further simplifying assumptions about the type of activity duration distributions. They also facilitate consideration of resource constraints expressed as quantity or time limits. Accuracy of estimates obtained by means of simulations can be improved by increasing the number of replications or by applying variance reduction methods. This paper analyzes how the method of variance reduction affects simulation results in terms of standard error of estimated project duration mean value. Three methods of reducing variance (Quasi Monte Carlo with sampling based on Weyl sequence, antithetic variates and Latin Hyper Cube Sampling) were examined for their efficiency in scheduling construction project under probabilistic conditions. The tests were conducted on a network model with the activity performance durations expressed with triangular distributions, which are applied in construction engineering for modeling the effects of variability of operating conditions. The tests assumed an availability limit on the workers employed at a construction site and time constraints for the initiation of specific construction processes. Results of simulation experiments are an indirect proof that applying variance reduction measures may reduce the time of the experiment as well as improve confidence in estimates of large scale models' characteristics. |
