Page 6 - EngineerIT July 2022

FRONT COVER STORY

Advanced operating conditions

require processes all working

together in a complex dance –

from PID to MPC

usiness sustainability is dependent indeterminate delays, operators often get impatient and make further changes to the same
on optimising return on investment or other variables to rectify a specific situation. Unfortunately, these sudden changes often
B(ROI) from fit for purpose, value-driven result in out-of-control processes.
solutions. Due to the complexity of the
required control conditions, companies are Why focus on process objectives – the optimum operating envelope
starting to turn to model predictive control Process performance variance impacts profitability; it is crucial to understand the
(MPC) to assist in decision-making as more circumstances that affect exceptional and poor performance - interpretation and decision
solutions become available due to the impact making within the production process environment. Decision-making by operational
of artificial intelligence and machine learning. personnel, even though assisted by sophisticated control systems (DCS, PLC or APC), plays a
Historically, many process controllers significant role in complex processes. Operational personnel typically need to make decisions
tuned their processes using PID single input, related to reagent dosing rate, feed rates, level, temperature, flow, pH set points and machine
single output (SISO) systems; some would speed. These decisions are applied at a functional unit or machine level. Multiple functional
even argue the relevance of PID today. units or machines are combined or configured to complete the value-add or conversion
However, the multiple input, multiple output process.
(MIMO) capability of the MPC methodology is The performance of the production process (which includes the physical process and
just more advanced and reaches the desired production and the technical teams’ quality of decisions) is measured in terms of the financial
result faster. Furthermore, MPC should be performance across the entire process.
more effective than normal PID controls Due to the nature of complex processes, performance measurements are typically
as it uses algorithms to predict based on aggregated production numbers. Typical management performance
outcomes based on current The future of measurements include production cost per weight or volume, overall equipment
operating conditions and process control and effectiveness (OEE), contribution margin per weight or volume, yield and
effect changes to setpoints recovery. There is thus a disconnect between the production team and
to improve performance, automation depends on management. Management measures performance in monetary terms,
reduce fluctuations and model predictive control whereas the operation team deals with set points to manage process stability,
prevent future adverse supported by AI and availability, process balances or energy usage at the functional unit level.
process conditions. machine learning. When the production team receives feedback on its performance, it merely
Manufacturing plant serves as a recording of past performance.
processes are complex too. In addition to profit, there may be other, more granular objectives to achieve
A small change in one process the profits. These can include waste reduction, emission reduction, energy and water-
variable may result in a small or large step saving, extending life-of-mine or equipment life, complying with regulatory measures and
change in another, depending on the specific others. For each objective, there is an optimum operating envelope (OOE) of variables that
process. These inter-relationships are not will promote the best possible achievement of the objectives. Unfortunately, sometimes the
easy to understand, and operators often variables for different objectives may be in total conflict. For example, in the mining industry,
overcompensate, resulting in fluctuating a short-term focus on metal recovery by mining more high-grade ore and less low-grade ore
process conditions. In addition, changing will increase recovery (yield) but can significantly shorten the economic viability life of the
one or more variable setpoints in a process mine.
may not be immediately apparent. Cycle
time, residence time, parallel processing and High ROI depends on achieving an optimum balance
other factors can delay changes in process For businesses to navigate this complexity, they must understand the optimum operating
conditions. In addition, due to the complexity envelope of each objective and how these need to be balanced to ensure ideal business
of the process, the time for a process change results by attaining all objectives to a certain degree. As a result of the conflicting objective
to have an effect often varies because of variables, not all objectives may be achieved completely. As in any tug-of-war, the end-point
other process conditions and cannot be will move one way or the other. The trick is to ensure optimum balance, where no objective is
consistently predicted. wholly neglected. The only way to do this is to deeply understand the factors influencing each
When confronted with these objective to define the optimum operating envelope, eventually leading to a higher ROI.

EngineerIT | July 2022 | 4