A semantic systems engineering framework for zero-defect engineering and operations in the continuous process industries

Abstract

The on-going twin transition demands that the continuous process industry builds and operates their facilities in a more sustainable way. This change affects the entire supply-chain. The market demands new ways of engineering, procuring and constructing plants that assure quality at each step of the process. Petroleum and petrochemical producers must reduce their waste and environmental footprint and find ways of migrating to sustainable production. There is zero tolerance for waste, emissions or process malfunctions. Engineering contractors need to transfer their skills to new processes and produce series, non-custom facilities for new applications like offshore wind energy, modular production and industrial symbiosis. This is leading to a convergence in methods with discrete manufacturing, especially the automotive industries. In this climate, this sector can benefit from applying Zero-defect Manufacturing (ZDM) to both engineering design and operations. This work defines a framework for implementing ZDM in the process industry supply chain. The framework brings together modelling techniques and models from the following disciplines: system engineering, computer-aided process engineering, automation (especially Industry 4.0) and semantic technologies. These contributions are synthesised into an information fabric that allows engineering firms to work in new ways. Operators and contractors can use the fabric to move from document-driven engineering to data-based processes. The fabric captures requirements and intent in design so that facilities can be delivered and started-up and operated with zero defects in the design and construction. The information is also a vital support for safe and efficient operations and maintenance. We call this zero-defect O&M. The framework combines a systems engineering break-down of facilities, based on ISO/IEC81346, with implementation in SysML, with semantic interoperability frameworks from the process industries (ISO15926). We build upon and synthesise the results of recent standardization initiatives from the industry, notably CFIHOS, DEXPI and READI. We draw on results from process systems engineering, the OntoCAPE ontology and the CAPE-OPEN standards. The framework is illustrated by application to a non-proprietary process system, namely the Tennessee-Eastman process. This example is used to show the modelling approach and indicate how the fabric supports zero-defect practices.

Tool Chain to Extract and Contextualize Process Data for AI Applications

“…Computer aided engineering (CAE) software plays an important role in the design phase of chemical plants. As described in [10], the asset lifecycle involves of a variety of software tools from the early phases where the functional requirements [11] are defined, over the more concrete functional design phase and the detailed asset specification up to the assets in operation. The collection of data during these design phases is important and good data quality does have a huge impact, e.g., for modern applications using AI technologies [12].…”

Graph Learning in Machine‐Readable Plant Topology Data

“…The statistical branch of AI (machine learning that is creating knowledge based on (lots of) data), can make use of these standards by having a tool-independent format and structure (i.e., ontology) to represent the results of the model-building via ML. A recent synthesis of some of the standards mentioned above into a framework for zero-defect engineering is given by [13].…”

“…This is in contrast to what is shown on a P&ID that contains the means of performing these process functions (e.g., centrifugal pump (P&ID) performs pumping (PFD), mixer (P&ID) performs mixing (PFD), etc.). Of course, these elements of the different worlds (process vs. plant) are related but not the same; see [13] for more detailed elaboration.…”