The total cost of any project is the sum of the costs of its components. At NASA, these components are called the Work Breakdown Structure (WBS). There are eleven numbered elements at the top level of the NASA WBS, including 1: Project Management, 2: Systems Engineering, 4: Science/Technology, 5: Payload(s), and 6: Spacecraft. During the earliest phases of the project life cycle, the costs of only one or a few of these components are well constrained, and the cost of the other elements must be estimated in order to submit a proposal with a reasonable chance of success. In the absence of time, money, or the level of detail to produce grassroots estimates, cost estimators look to past missions to gain insight; they have traditionally averaged the percentage allocations for each WBS breakdown from previously flown missions and used them to predict allocations for the new project that they are trying to cost. At IEEE 2022, we presented a novel costing method to project costs for future MIDEX class missions; rather than relying on average percentages, we proposed a Dirichlet-distributed model, which aptly describes the uncertainty of resource allocation by capturing the correlation between components. Here, we expand on our work from last year by (a) applying similar methodologies to more mission classes other than MIDEX (Flagship, New Frontiers, Discovery, SMEX, etc.), (b) imputing missing data records using expert knowledge of how costs are bookkept, (c) expanding our communication to align with up-to-date Uncertainty Quantification standards (categorizing uncertainty as aleatoric or epistemic), and (d) generating an interactive web tool easily accessible and usable by systems engineers.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Aleatoric and Epistemic Uncertainty Quantification in Bayesian Dirichlet Cost Rules of Thumb


    Beteiligte:
    Fleischer, Sam (Autor:in) / Hooke, Melissa (Autor:in)


    Erscheinungsdatum :

    2023-03-04


    Format / Umfang :

    2203706 byte




    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch






    Bayesian Confidence Calibration for Epistemic Uncertainty Modelling

    Kuppers, Fabian / Kronenberger, Jan / Schneider, Jonas et al. | IEEE | 2021


    NICM Schedule and Cost Rules of Thumb

    Habib-Agahi, H. / Ball, G. / Fox, G. et al. | British Library Conference Proceedings | 2009