It provides a systematic approach to reconciling data input errors. Reconciliation entails distributing mass imbalance errors across streams, adjusting specific streams to achieve a close mass balance, and using site-wide tools to seamlessly close the balance across assets. 5 Additionally, the process digital twin enhances operational efficiency through various capabilities. First, it ensures a pre - cise elemental balance apart from mass considerations, pro- viding a comprehensive understanding of the hydrocarbon processes. Moreover, it contains details about plant and tank farm operations, offering insights into movements critical for effective management. As a result, the digital twin helps uncover gross errors early in the business process, ranging from data entry errors and instrument failures to missing movements. With automatic logic, it uses coke production and adapts to flow variance, ensuring uninterrupted produc - tion even during outages. Supply chain – LP model updates for robust planning In refinery and petrochemical complexes, LP models play a vital role in assessing crude selection, yields, and gross mar- gin via optimisation functions. Despite their utility, these lin- ear models often face challenges due to infrequent updates in sensitivities related to feed, severity, and product qualities. This discrepancy between model predictions and actual per- formance, particularly at the end of back casting, can nega- tively impact operational efficiency.4 Traditional LP models used for planning, scheduling, and optimising assets lack continuous validation. This deficiency, often performed by individuals, creates inaccuracies that contribute to suboptimal operations. To overcome these challenges, the digital twin continuously tracks asset per- formance. Thus, all stakeholders get a comprehensive view of asset performance, including optimum operating targets, enhanced scheduling, and inventory cost savings. Additionally, digital twins not only automate complex work processes such as kinetic model calibration and validation but also leverage AI and ML methods to automate work - flows, check the application’s health, validate AI recalibration recommendations, and validate the accuracy of the vectors. As shown in Figure 5 , the automated model maintenance tool determines when the model needs to be recalibrated and establishes protocols for validating the model. The result
Max. Jet ooding
Max. Downcomer backup
70
70
60
80
60
50
80
74.4 %
46.8 %
90
50
40
90
100
30
100
40
Figure 3 Intrinsic parameters monitoring
is ongoing health score tracking, data quality analysis, and actionable email alerts. Process optimisation: Bridging gaps and identifying opportunities Process optimisation can be achieved using a digital twin to identify gaps between actual and benchmark performance during plant operation or the design stage. The gaps are analysed for corrective actions such as changing operating parameters or modifying equipment, piping, or instrumenta- tion. Digital twin applications for process optimisation include: • What-if analysis, debottlenecking, and optimisation • Constraint management • Molecular management • Unit/equipment optimisation • Product blending and stream routings • Identify margin improvement opportunities • Screen opportunities • Continuously track benefits for each implemented opportunity. Based on the authors’ experiences, the digital twin of an integrated refinery and petrochemical complex with a multi- feed steam cracker complex helped identify operational improvement opportunities exceeding 100 million USD and Capex savings tipping 100 million USD during the design review of its configuration. Real-time optimisation: Dynamic control for operational excellence In traditional distributed control systems (DCS), the process parameter from specified boundaries is common. In APC,
CCR-Coke Ratio Key Ratio
Mass Balance
Raw mass imbalance using outage
Max. Reconciled in mass ow
1.25
0
5
1.2
6
1.3
2
-2
4
1.35
4
7
1.15
-4
3
1.3
-2.4 WT%
5.2 WT%
1.4
2
-6
8
6
1.1
1
-8
9
1.45
8
1.05
1
10
10
0
-10
1.5
Figure 4 Key performance indicators
61
PTQ Q1 2024
www.digitalrefining.com
Powered by FlippingBook