Unconstrained KPI Variables in MPC for Root Cause Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Analyzing the root causes of poor Key Performance Indicators (KPI) performance in industrial processes is challenging due to the complexity of closed-loop control systems and the lack of integration with Model Predictive Control (MPC) models, which complicates understanding and addressing changes in process variables over time.
Innovation Solution
A method is introduced to analyze KPI performance by providing a dynamic MPC process model that includes KPIs from the business monitoring system, estimating future trajectories and steady-state values, and identifying dynamic relationships between key plant operating variables, allowing for the identification of causes of performance issues within the MPC controller.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional process control systems are used to manage industrial equipment, then basic control functions are maintained, but the ability to analyze root causes of KPI performance issues and optimize for changing conditions is limited
Solution Approach 1:
The patent merges the MPC controller with the business KPI monitoring system by integrating KPIs as unconstrained dependent variables into the MPC model. This combination allows the control system to simultaneously manage process control and analyze business performance, eliminating the need for separate analysis systems and enabling root cause analysis of KPI issues through the unified model.
Solution Approach 2:
The MPC controller is enhanced to perform multiple functions: traditional process control plus KPI monitoring, prediction, and root cause analysis. By making KPIs part of the MPC model with unconstrained dependent variables, the system universally handles both control and business performance analysis, allowing one system to serve multiple purposes without requiring additional specialized equipment.
2Measurement precision
If KPIs are added as constrained dependent variables in MPC, then KPI control is enabled, but the ability to analyze performance trends and identify root causes is compromised due to control limits
Solution Approach 1:
Instead of controlling KPIs directly as dependent variables with constraints (the conventional approach), the patent inverts the approach by making KPIs unconstrained dependent variables that are predicted by the model. This allows the system to monitor and analyze KPI performance trends without the constraints interfering with the analysis, while still enabling control through the manipulated variables.
Solution Approach 2:
The patent introduces unconstrained dependent variables as intermediaries between the MPC model and business KPIs. These KPI UDVs serve as mediators that capture business performance metrics without being directly constrained by control limits, allowing for accurate performance trend analysis while maintaining the control functionality through the manipulated variables and control limits of the original process.
3Productivity
If dynamic relationships between plant variables and KPIs are established, then real-time performance analysis is enabled, but computational complexity increases
Solution Approach 1:
The patent performs preliminary action by pre-establishing the dynamic relationships between plant variables and KPIs within the MPC model structure. The model is configured in advance to predict KPI values based on manipulated variables and process state, so that during real-time operation, the system can quickly query the model for KPI predictions without performing complex computations at runtime. This pre-configured model structure enables real-time analysis while minimizing computational burden.
Data Source
AI summary
A method of Key Performance Indicator (KPI) performance analysis and a dynamic Model Predictive Control (MPC) process model for an industrial process including measured variables (MVs) and controlled variables (CVs) for an MPC controller are provided. The MPC process model includes at least one KPI that is also included in a business KPI monitoring system for the industrial process. A future trajectory of the KPI and a steady-state (SS) value for the KPI are estimated. The future trajectory and SS value are used for determining dynamic relationships between key plant operating variables selected from the CVs and MVs, and the KPI. A performance of the KPI is analyzed including identifying at least one cause of a problem in the performance or exceeding the performance during operation of the industrial process from the dynamic relationships and a current value for at least a portion of the MVs.