Increased value and trustworthiness from digital twins

Whereas digital twins will impact decisions from early design to decommissioning, the full value of the digital twin will only be capitalized if there is confidence that the digital twin will function as specified. Lack of trust will limit the use and value gained of the digital twin. Feedback from the market indicate that many digital twins fail to deliver value because the end users do not trust the output from them. The digital twin is perceived as a “black box” and with limited ability for the user to determine if the results are correct or accurate. 

There is currently no common agreed standard that a digital twin can be developed, delivered and operated according to. DNVGL, together with TechnipFMC, have responded to this challenge by establishing a recommended practice for qualification and quality assurance of digital twins. 

The upcoming DNVGL RP on qualification and assurance of digital twins (DNVGLRP-A204), due to be published in October 2020, is built on the philosophy from qualification of technology (DNVGLRP-A203) and applies the principles from the RPs for data quality (DNVGL-RP-0497) and assurance of data-driven models (DNVGL-RP-0510).

The objective of the new RP is to describe a structured and systematic process and set requirements for the qualificationand assurance of digital twins, with the aim of obtaining trustworthy output from it. The recommended practice provides a common framework for organizations to be able to develop, integrate, qualify and operate a digital twin.

Target users typically include:
• asset operators – who request that a system supplier delivers a digital twin along with the asset/system.
• system suppliers – who require a systematic approach to ensure and document that the digital twin will perform according to expectations
• sub-suppliers – who want to be able to deliver a qualified module to be integrated into a larger digital twin

In the RP, a digital twin is defined as a virtual representation of a system or asset, that calculates system states and makes system information available, through integrated models and data, with the purpose of providing decision support, over its lifecycle. As can be seen from the definition above, a central part in the methodology is that the development of a DT should serve a clear purpose in order to provide value – it should provide decision support. In order to meet this objective, the twin is broken down into manageable parts (functional elements) that supports specific decisions and where the criticality of the functional element in terms of decision support, will govern the amount and level of activities to assure
the trustworthiness of the twin.

Another central part in the methodology is that the digital twin must be trusted over time. This is achieved by specifying requirements to a self-diagnostic indicator (e.g. traffic light) that reports the trustworthiness of the results provided by the functional element. The quality indicator combines continuous assessments (automated) and periodic assessment (manual) of the input data quality, failure modes or asset modifications that may negatively affect the output from
the functional element.

Solving the digital trust challenge will be key to its adoption, acceleration to use at greater scale, and acceptance as an accurate, valuable and trusted technology. The use of industry best practices is an efficient way of creating trust and thereby unleash the true value of digital twin technology