A design-based study of data-driven asset management for Storm Surge Barriers

Abstract

Rijkswaterstaat’s Storm Surge Barriers are expensive assets providing economic value with the expectation of future returns. The maintenance is executed by the asset management department, which aims at maximizing operability without sacrificing safety and reliability. However, regular maintenance becomes more stringent as circumstances change, and a budget shortfall is expected in the same timeframe. These operational and economical interferences urge the asset management department to adapt using more data-driven approaches replacing (some) physical inspections and only conducting maintenance when needed. This so-called data-driven asset management has the potential to reduce costs and ensure reliability.
Despite the promise, there is no guidance for developing data-driven asset management. In this thesis, design principles are developed to assist in better decision-making using data-driven technologies, resulting in effective maintenance. These new design principles integrate three different areas into the current asset management: Digital Twins (DT), providing a virtual environment for safely testing in various scenarios, Predictive Maintenance (PdM) for confidently predicting a future asset failure and Data Governance (DG), ensuring the data quality for appropriate decision-making for the mentioned data-driven technologies. In other words, the goal of this research is to develop design principles that overcome real-world asset data-related problems during the implementation of DT, PdM and DG within the asset management context.

The qualitative design-based research methodology consists of the following steps: Identifying real-world problems, developing appropriate solutions, and then finalizing by prescribing actionable sentences, also known as design principles. This research design was demonstrated in a real-life case for the Storm Surge Barriers asset class at Rijkswaterstaat (RWS). Two data-collecting research methods were utilized. The first data collecting method was the literature study, which acquired information from 25+ scientific papers from three developing fields in the scientific literature: PdM, DT and DG. The second was a series of semi-structured interviews held with 33 RWS interviewees in different parts of the organization, collecting real-life asset data problems and the associated best practices. To guide the interviews across the data-driven developments and the current status of data governance within RWS, an interview protocol was constructed. The interviews results revealed contrasting views on (1) the SSB openness to new technology, (2) the alignment between asset management priorities and the data-driven technology, and (3) data access by third parties. These contrasting viewpoints unknowingly construct invisible (almost) impermeable walls between different layers in the RWS organization, which prevents knowledge spillovers, resulting in departments maturing at different rates and impeding the understanding and communication between them.
 
As part of the research design, four types of design principles were formulated. The first was the interview-derived design principles. The second type was the literature derived design principles acquired by the results from the literature study. After that, the interview and the literature-derived principles were combined into 'the refined design principles’. The last type is the data governance design principles, for which the problems were inspired by semi-structured interviews and are solely solved by the current data governance literature.

Thereafter, the refined and data governance design principles were tested within the context of the DT and the PdM by using data flow diagrams. The benefit of using the data flow diagrams is to test if the selected design principles improve the practices of asset management.
Consequently, 33 design principles were developed across the four types of design principles that prescribe guidance on how PdM, DT and DG contribute to developing data-driven asset management. From these, 4 were data governance design principles, and 12 were refined design principles. The refined design principles were not found in prior literature and could be further grouped into short-term and long-term relevance for asset management.
The novelty that these refined design principles bring is twofold, filling in two gaps in the literature. Firstly, It brings clear guidelines which were previously scattered and unclearly presented across the scientific literature. Secondly, these design principles are approached in a novel manner by combining the core insights from the literature and the inclusion of empirical best practices in a real-life case.

The results of implementing the relevant data governance and refined design principles into the AM are twofold. Firstly, the DeP’s enhanced the DT from primarily visualization into a more multifunctional usable digital asset for various asset management purposes. Secondly, implementing the same set of principles enhanced the PdM by including the most important factors to produce reliable automatic predictions to better manage the maintenance schedule.

The managerial recommendation is that the digital twins and predictive maintenance development need to develop hand in hand with data governance aspects and utilize a system to centralize the knowledge across all storm surge barriers, like periodic communities of practices. A lack of these two crucial developments will inhibit RWS long-term vision of transitioning from a pure civil organization to a hybrid organization: A synergy between civil engineering discipline and data-driven technologies.