Browsing by Author "Schwabe, Oliver"

Now showing 1 - 12 of 12
  • Thumbnail Image
    ItemOpen Access
    An approach for selecting cost estimation techniques for innovative high value manufacturing products
    (Elsevier, 2016-11-02) Schwabe, Oliver; Shehab, Essam; Erkoyuncu, John Ahmet
    This paper presents an approach for determining the most appropriate technique for cost estimation of innovative high value manufacturing products depending on the amount of prior data available. Case study data from the United States Scheduled Annual Summary Reports for the Joint Strike Fighter (1997-2010) is used to exemplify how, depending on the attributes of a priori data certain techniques for cost estimation are more suitable than others. The data attribute focused on is the computational complexity involved in identifying whether or not there are patterns suited for propagation. Computational complexity is calculated based upon established mathematical principles for pattern recognition which argue that at least 42 data sets are required for the application of standard regression analysis techniques. The paper proposes that below this threshold a generic dependency model and starting conditions should be used and iteratively adapted to the context. In the special case of having less than four datasets available it is suggested that no contemporary cost estimating techniques other than analogy or expert opinion are currently applicable and alternate techniques must be explored if more quantitative results are desired. By applying the mathematical principles of complexity groups the paper argues that when less than four consecutive datasets are available the principles of topological data analysis should be applied. The preconditions being that the cost variance of at least three cost variance types for one to three time discrete continuous intervals is available so that it can be quantified based upon its geometrical attributes, visualised as an n-dimensional point cloud and then evaluated based upon the symmetrical properties of the evolving shape. Further work is suggested to validate the provided decision-trees in cost estimation practice.
  • Thumbnail Image
    ItemOpen Access
    Data relating to: "Dynamic multistep uncertainty prediction in spatial geometry" (2020)
    (Cranfield University, 2021-02-12 11:48) Grenyer, Alex; Schwabe, Oliver; ahmet Erkoyuncu, John; Zhao, Yifan
    Excel file corresponding to training data and results in conference paper - applied in MATLABImages: Figures 1-4 as in conference paperVideo: 3D plot rotationVideo: Conference presentation
  • Thumbnail Image
    ItemOpen Access
    Data relating to: "Multistep prediction of dynamic uncertainty under limited data" (2022)
    (Cranfield University, 2022-01-24 12:52) Grenyer, Alex; Schwabe, Oliver; ahmet Erkoyuncu, John; Zhao, Yifan
    Excel file: Symmetry trends used to establish correlation factorExcel file: Forecast method comparisonPowerPoint file: Figures in manuscriptPowerPoint file: Figures from additional simulationsMATLAB files to run the app and readme txt with instructions
  • Thumbnail Image
    ItemOpen Access
    Dynamic multistep uncertainty prediction in spatial geometry
    (Elsevier, 2021-02-10) Grenyer, Alex; Schwabe, Oliver; Erkoyuncu, John Ahmet; Zhao, Yifan
    Maintenance procedures for complex engineering systems are increasingly determined by predictive algorithms based on historic data, experience and knowledge. Such data and knowledge is accompanied by varying degrees of uncertainty which impact equipment availability, turnaround time and unforeseen costs throughout the system life cycle. Once quantified, these uncertainties call for robust forecasting to facilitate dependable maintenance costing and ensure equipment availability. This paper builds on the theory of spatial geometry as a methodology to forecast uncertainty where available data is insufficient for the application of traditional statistical analysis. To ensure continuous forecast accuracy, a conceptual dynamic multistep prediction model is presented applying spatial geometry with long-short term memory (LSTM) neural networks. Based in MATLAB, this deep learning model predicts uncertainty for the in-service life of a given system. The further into the future the model predicts, the lower the confidence in the uncertainty prediction. Forecasts are therefore also made for a single time step ahead. When this single step is reached in real time, the next step is forecast and used to update the long range prediction. The uncertainty here is contributed by an aggregation of quantitative data and qualitative, subjective expert opinions and additional traits such as environmental conditions. It is therefore beneficial to indicate which of these factors prompts the greatest impact on the aggregated uncertainty for each forecast point. Future work will include the option to simulate and interpolate input data to enhance the accuracy of the LSTM and explore suitable approaches to mitigate, tolerate or exploit uncertainty through deep learning.
  • Thumbnail Image
    ItemOpen Access
    Dynamics of cost uncertainty for innovative high value manufacturing products - a geometric phenomenon
    (Elsevier, 2018-11-24) Schwabe, Oliver; Shehab, Essam; Erkoyuncu, John Ahmet
    In practice the forecasting of cost uncertainty for high value manufacturing products is typically a statistical exercise focused on predicting a static cost range at a future point in time. This only leads to robust forecasts if sufficient historical data is available, robust knowledge of cost estimating relationships exists and these relationships do not change in the time between creating the forecast and verifying its accuracy. The more innovative the product is the less likely it however is that these prerequisites are met. Using cost data from the U.K. Ministry of Defence Royal Air Force A400M transport aircraft from 2002 to 2014 as an example, the dynamics of cost estimating relationships over time are examined using a novel non-statistical forecasting approach. The approach considers cost uncertainty as a geometric phenomenon, does not rely on prior information and permits easy identification of patterns in changes of cost estimating relationships over time.
  • Thumbnail Image
    ItemOpen Access
    A framework for geometric quantification and forecasting of cost uncertainty for aerospace innovations
    (Elsevier, 2016-06-16) Schwabe, Oliver; Shehab, Essam; Erkoyuncu, John Ahmet
    Quantification and forecasting of cost uncertainty for aerospace innovations is challenged by conditions of small data which arises out of having few measurement points, little prior experience, unknown history, low data quality, and conditions of deep uncertainty. Literature research suggests that no frameworks exist which specifically address cost estimation under such conditions. In order to provide contemporary cost estimating techniques with an innovative perspective for addressing such challenges a framework based on the principles of spatial geometry is described. The framework consists of a method for visualising cost uncertainty and a dependency model for quantifying and forecasting cost uncertainty. Cost uncertainty is declared to represent manifested and unintended future cost variance with a probability of 100% and an unknown quantity and innovative starting conditions considered to exist when no verified and accurate cost model is available. The shape of data is used as an organising principle and the attribute of geometrical symmetry of cost variance point clouds used for the quantification of cost uncertainty. The results of the investigation suggest that the uncertainty of a cost estimate at any future point in time may be determined by the geometric symmetry of the cost variance data in its point cloud form at the time of estimation. Recommendations for future research include using the framework to determine the “most likely values” of estimates in Monte Carlo simulations and generalising the dependency model introduced. Future work is also recommended to reduce the framework limitations noted.
  • Thumbnail Image
    ItemOpen Access
    A geometrical framework for forecasting cost uncertainty in innovative high value manufacturing.
    (2018-05) Schwabe, Oliver; Shehab, Essam; Erkoyuncu, John Ahmet
    Increasing competition and regulation are raising the pressure on manufacturing organisations to innovate their products. Innovation is fraught by significant uncertainty of whole product life cycle costs and this can lead to hesitance in investing which may result in a loss of competitive advantage. Innovative products exist when the minimum information for creating accurate cost models through contemporary forecasting methods does not exist. The scientific research challenge is that there are no forecasting methods available where cost data from only one time period suffices for their application. The aim of this research study was to develop a framework for forecasting cost uncertainty using cost data from only one time period. The developed framework consists of components that prepare minimum information for conversion into a future uncertainty range, forecast a future uncertainty range, and propagate the uncertainty range over time. The uncertainty range is represented as a vector space representing the state space of actual cost variance for 3 to n reasons, the dimensionality of that space is reduced through vector addition and a series of basic operators is applied to the aggregated vector in order to create a future state space of probable cost variance. The framework was validated through three case studies drawn from the United States Department of Defense. The novelty of the framework is found in the use of geometry to increase the amount of insights drawn from the cost data from only one time period and the propagation of cost uncertainty based on the geometric shape of uncertainty ranges. In order to demonstrate its benefits to industry, the framework was implemented at an aerospace manufacturing company for identifying potentially inaccurate cost estimates in early stages of the whole product life cycle.
  • Thumbnail Image
    ItemOpen Access
    A maturity model for rapid diffusion of innovation in high value manufacturing
    (Elsevier, 2021-02-10) Schwabe, Oliver; Bilge, Pinar; Hoessler, Andreas; Tunc, Taner; Gaspar, Daniel; Price, Nigel; Sharir, Lee; Pasher, Edna; Erkoyuncu, John Ahmet; de Almeida, Nuno Marques; Formica, Piero; Schneider, Lynne; Dietrich, Franz; Shehab, Essam
    In order to support accelerating the diffusion of innovations in high value manufacturing related to enabling flexible mass customization, this paper presents a research-based maturity model for forecasting the speed of innovation diffusion from ideation to market saturation. The model provides an early stage applied research view of (groups of) “game changing” variables, which accelerate diffusion of innovations to significantly reduce financial uncertainty and minimize the time to derive value from the original idea. The model is applied to multiple case studies related to the repurposing and customization of existing mass manufacturing infrastructures and processes to meet novel requirements. Case studies include among others a reference model based on a literature review, the diffusion of 3-D printing technology in manufacturing, the diffusion of novel cement manufacturing technology and the manufacturing of intensive care ventilators during the Covid-19 pandemic. The diffusion of innovation model applied is based on diffusion of innovation principles founded in the research of Everett Rodgers, the Bass Diffusion Curve and aligned to recent advances in living (eco-) systems theory. Special emphasis is placed on determining not only the relevance of “known-known” success factors for rapid innovation diffusion, but also on identifying “unknown-unknown” game changers enabling the required changes at pace. Key findings are that “game changing” factors for the innovations are primarily the interdependent availability of budget and resources to achieve market saturation, urgency of need shared by all participants, observability of impact (value creation) and compatibility with existing ways of work. Critical as well is population of all diffusion web roles with unique individuals. Further research is suggested regarding the dependency of assessed variable (groups) and the integration of Technical Readiness Level phases into the forecasting model.
  • Thumbnail Image
    ItemOpen Access
    Multistep prediction of dynamic uncertainty under limited data
    (Elsevier, 2022-01-12) Grenyer, Alex; Schwabe, Oliver; Erkoyuncu, John Ahmet; Zhao, Yifan
    Engineering systems are growing in complexity, requiring increasingly intelligent and flexible methods to account for and predict uncertainties in service. This paper presents a framework for dynamic uncertainty prediction under limited data (UPLD). Spatial geometry is incorporated with LSTM networks to enable real-time multistep prediction of quantitative and qualitative uncertainty over time. Validation is achieved through two case studies. Results demonstrate robust prediction of trends in limited and dynamic uncertainty data with parallel determination of geometric symmetry at each time unit. Future work is recommended to explore alternative network architectures suited to limited data scenarios.
  • Thumbnail Image
    ItemOpen Access
    On the change of cost risk and uncertainty throughout the life cycle of manufacturing products
    (2020-02-18) Schwabe, Oliver; Erkoyuncu, John Ahmet; Shehab, Essam
    In practice cost estimators typically assume that cost risk and uncertainty continuously decrease across the whole product life cycle. Industry case studies and semi-structured interviews indicate that while cost risk and uncertainty decreases between technology readiness levels / stage gates, it increases when technology readiness levels / stage gates change. This increase can lead to cost risk and uncertainty levels above those at previous technology readiness levels / stage gates. This difference between assumptions in practice and evidence from case studies and semi-structured interviews may lead to the over- and / or under-assignment of capital reserves over time, thus resulting in binding project capital unnecessarily and / or the need to increase projects budgets in an unplanned manner. Further research is suggested regarding the scale of changes in cost risk and uncertainty when technology readiness level changes / stage gates are arrived at in order to improve robustness of forecasting efforts
  • Thumbnail Image
    ItemOpen Access
    Short interval control for the cost estimate baseline of novel high value manufacturing products – a complexity based approach
    (Elsevier, 2016-11-02) Schwabe, Oliver; Shehab, Essam; Erkoyuncu, John Ahmet
    Novel high value manufacturing products by default lack the minimum a priori data needed for forecasting cost variance over of time using regression based techniques. Forecasts which attempt to achieve this therefore suffer from significant variance which in turn places significant strain on budgetary assumptions and financial planning. The authors argue that for novel high value manufacturing products short interval control through continuous revision is necessary until the context of the baseline estimate stabilises sufficiently for extending the time intervals for revision. Case study data from the United States Department of Defence Scheduled Annual Summary Reports (1986-2013) is used to exemplify the approach. In this respect it must be remembered that the context of a baseline cost estimate is subject to a large number of assumptions regarding future plausible scenarios, the probability of such scenarios, and various requirements related to such. These assumptions change over time and the degree of their change is indicated by the extent that cost variance follows a forecast propagation curve that has been defined in advance. The presented approach determines the stability of this context by calculating the effort required to identify a propagation pattern for cost variance using the principles of Kolmogorov complexity. Only when that effort remains stable over a sufficient period of time can the revision periods for the cost estimate baseline be changed from continuous to discrete time intervals. The practical implication of the presented approach for novel high value manufacturing products is that attention is shifted from the bottom up or parametric estimation activity to the continuous management of the context for that cost estimate itself. This in turn enables a faster and more sustainable stabilisation of the estimating context which then creates the conditions for reducing cost estimate uncertainty in an actionable and timely manner.
  • Thumbnail Image
    ItemOpen Access
    Uncertainty quantification metrics for whole product life cycle cost estimates in aerospace innovation
    (Elsevier, 2015-06-25) Schwabe, Oliver; Shehab, Essam; Erkoyuncu, John Ahmet
    The lack of defensible methods for quantifying cost estimate uncertainty over the whole product life cycle of aerospace innovations such as propulsion systems or airframes poses a significant challenge to the creation of accurate and defensible cost estimates. Based on the axiomatic definition of uncertainty as the actual prediction error of the cost estimate, this paper provides a comprehensive overview of metrics used for the uncertainty quantification of cost estimates based on a literature review, an evaluation of publicly funded projects such as part of the CORDIS or Horizon 2020 programs, and an analysis of established approaches used by organizations such NASA, the U.S. Department of Defence, the ESA, and various commercial companies. The metrics are categorized based on their foundational character (foundations), their use in practice (state-of-practice), their availability for practice (state-of-art) and those suggested for future exploration (state-of-future). Insights gained were that a variety of uncertainty quantification metrics exist whose suitability depends on the volatility of available relevant information, as defined by technical and cost readiness level, and the number of whole product life cycle phases the estimate is intended to be valid for. Information volatility and number of whole product life cycle phases can hereby be considered as defining multi-dimensional probability fields admitting various uncertainty quantification metric families with identifiable thresholds for transitioning between them. The key research gaps identified were the lacking guidance grounded in theory for the selection of uncertainty quantification metrics and lacking practical alternatives to metrics based on the Central Limit Theorem. An innovative uncertainty quantification framework consisting of; a set-theory based typology, a data library, a classification system, and a corresponding input-output model are put forward to address this research gap as the basis for future work in this field.