#34 - Chat with Michel Ruel
I had the pleasure of chatting with Michel Ruel, a Canadian electrical engineer, Fellow of the International Society of Automation (ISA), and a legend in the process control industry. With over forty years of experience, Michel has worked in various fields, including pulp and paper, mining, and petrochemical, tackling some very challenging problems with great return on investment to his clients. During our conversation, we explored his professional journey, his numerous projects, and his significant contributions to the field.
Michel started his career in the 1970s at a pulp mill plant, which is a great place for a process control engineer to learn. These plants have many complex control loops, like those for water, brightness, stock flow, blend chest consistency, and weight. These systems often interact, causing oscillations, and the harsh conditions of the process push the sensors and actuators to their limits, making it a very challenging environment.
Michel studied process control in a technical college (technician) then got an engineering degree (engineer). He is pleased to explain that all over his career, he used those two “hats” to optimise processes.
After about twenty years of gaining experience and working on various projects, along with teaching at local colleges and universities, Michel decided to start his own consulting and engineering company with a colleague. By then, he had built a strong reputation and knew many superintendents and plant directors interested in his services. He quickly formed a team of specialists who, like special forces, were well-trained, mission-focused, and capable of handling highly critical projects, all around the world, that others had either failed at or refused to take on.
Michel's approach is very pragmatic. This is evident from his many articles, his classes, his books and conversations with him. He firmly believes that tuning the loop should be the last step for a process control engineer. Tuning should only occur after thoroughly understanding the system and the plant. Additionally, he emphasises that his main goal is to maximise the use of the plant's existing assets to bring the highest profits to his clients. His projects, often with budgets of several hundred thousand dollars, typically achieved a return on investment (ROI) in weeks or a few months, rather than years. This is not only a simple detail. Michel has personally seen how this rapid ROI saved companies from bankruptcy in some cases.
When starting a project, Michel prefers to visit the field to observe the process and equipment first-hand and to talk with the operators on the factory floor and in control rooms; this is what he calls “walk the process”. This approach helps him fully understand how the process should work and identify the main issues and potential opportunities. After this initial stage, Michel begins collecting process data, such as the control loop's set-point (SP), process variable (PV), controlled value (CV), and mode (manual/auto). This data is typically collected at the same sample rate as the loop's scan time.
The second stage of Michel's study involves a deeper analysis of additional parameters of the target control loops, such as dead-time, process variable filter, integral time, and derivative time. Combining his extensive experience and an analytical approach, Michel has developed a decision tree model to help his team determine the best control strategy. This model guides them in choosing the most appropriate approach, ranging from basic PID control to more advanced methods such as cascade or feed-forward control (ARC advanced regulatory control), model predictive control (MPC), fuzzy logic control (FLC), and neural networks (AI).
One notable success case occurred in an application in the mining industry, involving a Semi-Autogenous Grinding (SAG) Mill. These mills are large cylindrical vessels that rotate, using their movement and mechanical design to grind ore into smaller particles. From Michel’s paper, “the SAG mill stability had to be improved and throughput increased. The process is multivariable, strongly non-linear, and before implementing this system, the operators were actively manipulating many variables with varying success depending on operator experience and occurring disturbances.” Mimicking an experienced operator orchestrating a complex process is a typical application for a fuzzy logic control strategy, particularly in the mining industry where it is very hard to accurately model the process and its disturbances.
This approach is quite uncommon in the industry, and Michel found inspiration for it after attending a technical conference in San Francisco. Sometimes, this kind of out-of-the-box thinking is necessary to effectively solve a problem. A fuzzy logic controller typically involves three steps. The first is fuzzification, where input parameters are selected and categorised into groups such as Low (L), Low-Low (LL), High (H), and High-High (HH). In this case, the input variables were Load, Load rate, Ore Size, Ore Size rate, Recirculation, Recirculation rate, Power, Power rate, Density, and Density rate. The second step involves creating and evaluating hundreds of process rules based on these input values and their combinations. Finally, in the defuzzification step, the final outputs are calculated, which in this example were Ore tonnage, Water flow, and Rotation speed.
Since his retirement a few years ago, Michel has continued to contribute to the community. For example, last year saw the release of the ISA-TR5.9-2023, a master class technical report on PID Algorithms and Performance, which took about three years to develop and in which he served as co-chair. He still provides training on advanced control and actively shares his knowledge through writing articles and mentoring the younger generation through programs such as “Ask the Automation Pros”. Reflecting on his career, he recognizes a significant skill shortage in this field, which is crucial to many industries.
Links: