If you could ask a plant only one question about the health of its base-layer control, a good one would be: how many of your controllers are switched off? Percent of loops in manual answers that, counting the share of controllers that operators have taken out of automatic and are running by hand. It is often the single most telling number in a loop assessment, because a loop in manual is automation that has been abandoned. This guide explains what the metric means, why operators end up flipping loops to manual, the hidden cost of all that idle automation, and how monitoring mode tags across a fleet surfaces the trend.
Percent of Loops in Manual in one line: Percent of loops in manual is the fraction of a plant's control loops that operators have switched from automatic to manual mode, running them by hand instead of letting the controller regulate. It is a plant-level diagnostic and often the most revealing single number in a loop assessment, because every loop in manual is a piece of automation that is not being used. High percentages usually mean loops are misbehaving, from bad tuning, valve stiction, wrong valve action, or nuisance alarms, so operators bypass them, quietly surrendering the energy, throughput, and stability that automatic control was meant to deliver.
A control loop exists to do a job automatically, holding a flow, a level, a pressure, or a temperature at setpoint without a human babysitting it. When an operator switches that loop to manual, the automation stops and a person is now setting the output by hand, adjusting it only when they happen to notice a problem and have the attention to spare. Percent of loops in manual counts how widespread that surrender is across the whole plant, and because it rolls up every abandoned loop into one figure, it captures the overall state of base-layer automation in a way no single loop ever could.
The reason it is so telling is that it reflects real human judgement, not a theoretical performance calculation. Operators are practical people who keep loops in automatic when they trust them and flip them to manual when they do not, so a high percent in manual is a plant full of loops that the people closest to them have decided are not worth trusting. That collective verdict is hard to argue with and points straight at a problem that abstract metrics can miss: the automation is there, it cost money to install, and it is sitting unused because it does not work well enough to leave alone.
As a diagnostic, it is also refreshingly concrete. You do not need to interpret an index or a spectrum; you simply count controllers in manual and divide by the total. That plainness makes it a natural headline number for a loop assessment, understandable by operators, engineers, and managers alike, and a good starting point before diving into the finer measures of how well the loops still in automatic are actually controlling.
Operators rarely take a loop to manual out of preference; they do it because the loop in automatic is causing them grief. The most common reason is poor tuning, where a loop reacts too aggressively and oscillates or hunts, disturbing the process, so the operator takes over to steady it. Another frequent culprit is valve stiction, a sticky control valve that does not move smoothly, so the loop cannot settle and the operator gives up on automatic and nudges the valve manually. A valve configured with the wrong action, or an undersized valve that saturates, produces the same result: a loop that fights the operator until they switch it off.
Sometimes the reason is not the loop's control quality at all but the noise around it. A loop that triggers nuisance alarms whenever it is in automatic may be flipped to manual purely to silence the alarms, trading control for quiet. Loops are also left in manual after an upset or a startup and simply never returned to automatic, because nobody had a reason or a reminder to switch them back, so the manual state persists long after the situation that caused it has passed.
What these causes share is that manual mode is a symptom, not a root cause. The operator flipping the switch is responding rationally to a loop that is failing them, so the number of loops in manual is really a count of underlying faults, sticky valves, bad tuning, wrong valve action, nuisance alarms, that have driven people to bypass the automation. This is why raising the metric by simply forcing loops back to automatic does not work: without fixing what drove operators to manual, they will switch the loops off again, and the percent in manual will climb back up.
The cost of loops in manual is real but easy to overlook because nothing dramatic breaks. A loop in manual is controlled only as well and as often as a busy operator can attend to it, which is far less precisely than an automatic controller running continuously. That looseness shows up as wider swings around the ideal operating point, which waste energy, let product quality drift, push equipment harder, and leave throughput on the table, all quietly, all the time. Because no single loop in manual causes a visible failure, the aggregate cost of many loops in manual accumulates unnoticed across the plant.
There is also an opportunity cost in the automation itself. Every loop represents an investment in instrumentation, valves, and control configuration made precisely so a human would not have to regulate that variable by hand. A loop sitting in manual writes off that investment, turning a designed-in efficiency into an unused asset while adding to the operator's manual workload. A plant with many loops in manual is both paying for automation it is not using and loading its operators with work the automation was supposed to remove.
Monitoring mode tags across a fleet is how the metric becomes actionable rather than anecdotal. Because every loop reports its mode, and a cloud SCADA platform historises those mode tags, the percent in manual can be computed continuously for a unit, a site, or an entire fleet, and watched as a trend. A platform such as Merobix can surface which sites carry the most loops in manual and whether the number is rising, turning a vague sense that operators bypass a lot of loops into a tracked figure that maintenance and engineering can drive down by fixing the specific faults behind each abandoned loop.
Because every loop in manual is automation that has been abandoned, so the metric captures, in one figure, how much of the plant's base-layer control is actually being used. It reflects the practical judgement of operators, who keep loops in automatic when they trust them and flip them to manual when they do not, making it a hard-to-argue-with verdict on control health. It is also simple to compute and to communicate, which is why it often headlines a loop assessment.
Almost always because the loop misbehaves in automatic and manual is the path of least resistance. Common causes are poor tuning that makes the loop oscillate, a sticky valve that stops the loop settling, a wrongly configured or undersized valve that saturates, and nuisance alarms the operator silences by taking over. Loops are also sometimes left in manual after an upset or startup and never switched back. In every case, manual mode is a symptom of an underlying fault, not the real problem.
By fixing the faults that drove operators to manual in the first place, not by simply forcing loops back to automatic, which just gets undone. That means retuning oscillating loops, repairing or replacing sticky valves, correcting wrong valve action, and quieting nuisance alarms, so operators trust the loops enough to leave them in automatic. Monitoring mode tags across the fleet identifies which loops and sites are worst, so the remediation effort targets the specific causes behind each abandoned loop.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.