What the data for this category looks like
Core data for petroleum refining scenarios originates from production operations, process, and scheduling departments of oil refining enterprises. It covers documents including operating procedures, production plans, material balances, distillation outlet quality reports, interlock ledgers, shift handover records, and change management documents. Data sources include DCS/APC systems, LIMS laboratory management systems, tank farm metering systems, crude oil evaluation databases, equipment management systems, and energy systems. Data update cycles include real-time (unit operation parameters), shift-based (shift handover records), and daily (material balance reports). Fields include crude oil API gravity, distillation outlet temperature, sulfur content, equipment interlock status, and other metrics, with corresponding physical units attached.
How These Characteristics Create Constraints for Workflow Orchestration
Differentiated update cycles across multiple data sources require workflow configurations to align with trigger timings. Real-time data must be bound to real-time trigger nodes. Shift-based and daily data require scheduled trigger tasks. The coupled logistics and energy network across multiple continuous units requires workflows to pass cross-unit parameters via variables to prevent data inconsistency between nodes. Diverse document types require dedicated parsing rules for different document categories such as operating procedures and quality reports, to ensure accurate extraction of audit fields. Large fluctuations in raw material properties require workflow configuration of branch routing rules, to adjust audit logic based on crude oil properties. Businesses with high-severity error consequences require retry mechanisms and alert nodes, to block abnormal processes in a timely manner.
Configuration Settings
This chapter’s configurations apply to FastGPT v4.17.0, and are adapted to the workflow orchestration needs of petroleum refining scenarios.
| Configuration Item | Recommended Setting Method | Basis for This Setting |
|---|---|---|
| Trigger Timing | Bind corresponding nodes separately for real-time data updates, shift-based scheduling, and daily scheduling | Match the update frequencies of different data types |
| Timeout Setting | 600 seconds | Crude oil evaluation and material balance data processing takes a long time |
| Branch Routing Rules | Divide branches based on crude oil API gravity and sulfur content ranges | Adapt audit logic for crude oil property fluctuations |
| Node Retry Count | 3 times | Address temporary execution exceptions caused by process fluctuations |
variables Mapping Rules | Map LIMS quality data and DCS operation parameters to audit variables | Unify audit indicators across data sources |
nodeId Binding | Configure unique identifiers for each unit scheduling node | Distinguish execution nodes across multiple workflows |
Three Common Configuration Mistakes
- Phenomenon: A "duplicate node" error is returned when calling multiple workflows. Cause: Unique identifiers are not configured for the
nodeIdof different workflows. - Phenomenon: Knowledge base recall results during workflow execution do not match audit requirements. Cause:
variablesare not bound to crude oil property and quality indicator parameters. - Phenomenon: Exported workflows cannot be directly imported into third-party integration platforms. Cause: The exported configuration file is not sorted according to the
nodeIdorder.
How to Verify a Complete Configuration
- Check whether the
variablesmapping covers all quality indicators and operation parameter fields required for audits. - Verify whether the nodes bound by
nodeIdcover the scheduling processes and audit links of all associated units. - Test whether workflows with different trigger timings match the update cycles of corresponding data.
- Simulate process fluctuation scenarios to confirm that branch routing rules correctly trigger corresponding audit logic.
The values provided here are common starting points and should be evaluated against the reader's own samples.