TY - THES

T1 - Best Practices and Lessons Learnt of Probabilistic Schedule Analysis for Oil and Gas Field Development Projects

AU - Katschnig, Matthias

N1 - embargoed until null

PY - 2015

Y1 - 2015

N2 - It is an axiom of economically act that future is uncertain. This fact is known in particular by the oilfield industry which naturally is an industry that faces many uncertainties (un-known variables) and risks. Especially project scheduling is concerned by this problem, because its job is to make future events manageable. Thus the analysis of project schedules is an important part of project management in order to manage project schedule risk adequately. This procedure is called probability schedule analysis (PSA). Key issues of PSA are estimating the project durations (and consecutively costs), finding the most critical risks and important impact factors and determining mitigating or avoiding strategies. Therefore, the core of this study was to collect best practices of probabilistic schedule analysis from the theoretical (literature) view and conduct some PSA examples with @Risk and MS Project from the practical point of view. To do so this work has four base chapters (Chapter 1: theoretical probability schedule analysis, Chapter 2: input data estimation, Chapter 3: company survey, Chapter 4: applied probability schedule analysis) studying the most important fields of today`s PSA. The literature outcome says that probabilistic schedule analysis is superior to a deterministic approach by taking uncertainty and therefore reality into account. In many cases Monte Carlo simulation was and is used. On the other hand the more straightforward PERT analysis can give you a raw and quick approach to possible values. Furthermore collecting the right input data is more crucial than finding a “magical input distribution”. Input data estimation for your schedule analysis must include diverse groups of estimators (controlled by some estimation workshop) which are aware of black swans, cognitive biases, heuristics and the statistical concept of crowd wisdom. The company survey basically confirms the literature output, moreover a transparent and clearly communicated PSA guideline is demanded and PSA outcomes must be well documented for presentation. On the practical side @Risk simulation results state that the main key factors are input distribution shape, input distribution mean and spread, the Central Limit Theorem, task constraints and task correlations. All of them have specific impacts on the completion date of the project and should be carefully revisited during scheduling. All results mentioned above are implicated in a flowchart suggesting a way of doing PSA. Firstly an estimation workshop takes place where probabilistic input data is generated. Ideally this is combined with producing the deterministic baseline schedule, thus the same people can work on both parts. As result a probabilistic schedule is achieved that can enter the Monte Carlo simulation stage. Schedule risk drivers are now (hopefully) detected and schedule can be optimised. One important process step is filling, checking and maintaining a Monte Carlo input value database. This database builds the foundation for subsequent estimation workshop and will be an assessment reference in the following process steps. Overall the true intent of PSA is encompassing all uncertainties to elicit confidence inter-vals in order to make better decisions and highlight important duration and cost drivers.

AB - It is an axiom of economically act that future is uncertain. This fact is known in particular by the oilfield industry which naturally is an industry that faces many uncertainties (un-known variables) and risks. Especially project scheduling is concerned by this problem, because its job is to make future events manageable. Thus the analysis of project schedules is an important part of project management in order to manage project schedule risk adequately. This procedure is called probability schedule analysis (PSA). Key issues of PSA are estimating the project durations (and consecutively costs), finding the most critical risks and important impact factors and determining mitigating or avoiding strategies. Therefore, the core of this study was to collect best practices of probabilistic schedule analysis from the theoretical (literature) view and conduct some PSA examples with @Risk and MS Project from the practical point of view. To do so this work has four base chapters (Chapter 1: theoretical probability schedule analysis, Chapter 2: input data estimation, Chapter 3: company survey, Chapter 4: applied probability schedule analysis) studying the most important fields of today`s PSA. The literature outcome says that probabilistic schedule analysis is superior to a deterministic approach by taking uncertainty and therefore reality into account. In many cases Monte Carlo simulation was and is used. On the other hand the more straightforward PERT analysis can give you a raw and quick approach to possible values. Furthermore collecting the right input data is more crucial than finding a “magical input distribution”. Input data estimation for your schedule analysis must include diverse groups of estimators (controlled by some estimation workshop) which are aware of black swans, cognitive biases, heuristics and the statistical concept of crowd wisdom. The company survey basically confirms the literature output, moreover a transparent and clearly communicated PSA guideline is demanded and PSA outcomes must be well documented for presentation. On the practical side @Risk simulation results state that the main key factors are input distribution shape, input distribution mean and spread, the Central Limit Theorem, task constraints and task correlations. All of them have specific impacts on the completion date of the project and should be carefully revisited during scheduling. All results mentioned above are implicated in a flowchart suggesting a way of doing PSA. Firstly an estimation workshop takes place where probabilistic input data is generated. Ideally this is combined with producing the deterministic baseline schedule, thus the same people can work on both parts. As result a probabilistic schedule is achieved that can enter the Monte Carlo simulation stage. Schedule risk drivers are now (hopefully) detected and schedule can be optimised. One important process step is filling, checking and maintaining a Monte Carlo input value database. This database builds the foundation for subsequent estimation workshop and will be an assessment reference in the following process steps. Overall the true intent of PSA is encompassing all uncertainties to elicit confidence inter-vals in order to make better decisions and highlight important duration and cost drivers.

KW - Theoretische Zeitplananalyse

KW - Praktische Zeitplananalyse

KW - Unternehmensumfrage

KW - Modell für Kosten-/Zeitpläne

KW - theoretical probalistic schedule analysis

KW - data estimation workhop

KW - company survey

KW - applied probabilistic schedule analysis

KW - model for propabilistic analysis (costs/schedule)

M3 - Master's Thesis

CY - Leoben

ER -