The world’s first deep reinforcement learning solution for hydrocarbon processors
Your most profitable processes are your hardest—and our specialty
Process optimization is complex
Refineries and chemical plants are complicated to model, control, and optimize. There are thousands of interdependent process variables requiring recalibration on a continuous 24/7 basis. These thousands of variables are continuously disturbed by ever-varying feedstock composition, ambient conditions, utilities, equipment status, and product prices. The interdependency of the relationships between all of these variables makes it hard to keep the process optimized at all times.
The traditional process control and optimization hierarchy is limited
Over the past few decades, our entire industry has grown accustomed to operating within a hierarchy of layers, each with their own engineering discipline, methodologies, modeling techniques, and team. While this hierarchical control and optimization stack stabilizes plant operations, it often overlooks a small number of pivotal process variables. These key variables, when identified and optimized continuously, can generate millions in additional annual margin on each major refining process unit. This margin often is missed because the siloed layers cannot keep the key variables in your most complicated units optimized every minute of every day. Optimizing these key variables would require a single model to simultaneously understand the minute-by-minute resolution of the process control layers, as well as the global and plant-wide planning and economics layers.
Closed Loop Neural Network™
for interconnected process optimization
Capturing millions in additional annual margin is possible, but requires a new approach and new technology. Simply bolting machine learning or AI solutions onto specific layers may help with engineering productivity, but will not provide an optimization step-change and will not add millions in new margins to your bottom line.
Imubit’s Closed Loop Neural Network™ operates across all of the hierarchical layers and interconnects the different teams. It captures nonlinear dynamics, uses subtle data patterns to compensate for missing composition measurements, and directly manipulates process control variables to optimize your most complicated processes.
What does this mean for you?
A seamless, interconnected flow of data and communication that allows you to optimize process across multiple units, in real time, and unlock hidden margins within your plant.
Imubit Presents at the AFPM Summit 2021
In October 2021, the American Fuel & Petrochemical Manufacturers (AFPM) invited Imubit CEO Gil Cohen to participate in a series of panels discussing AI in refining, and advances in APC. You can learn more about Imubit and the world’s first deep reinforcement learning solution for hydrocarbon processors by watching the presentation videos below.