Before you can tune a loop or interpret its behavior, you have to know what kind of process you are dealing with, and the most fundamental split is between self-regulating and integrating processes. A self-regulating process finds its own new resting point after a change; an integrating one just keeps going, ramping without limit until something stops it. This single distinction governs which tuning rules apply, how a bump test is read, and why level loops behave so differently from flow loops. This guide explains both types, the tell that separates them, and why getting the classification wrong is a classic tuning mistake.
Integrating vs Self-Regulating in one line: A self-regulating process settles to a new steady state on its own after a step change in the controller output, like flow or temperature reaching a new stable value. An integrating, or non-self-regulating, process instead ramps continuously and never settles by itself, like the level in a tank or the pressure in a closed vessel, where a mismatch between in and out accumulates without bound. The two types respond differently to a bump test and require different tuning approaches, so identifying which one a loop is comes before any tuning.
A self-regulating process has a built-in balancing tendency. Step the controller output and the process moves, but as it moves it changes the balance of forces acting on it in a way that opposes further change, so it eases toward a new equilibrium and stops there. Flow is the clearest example: open a valve a bit more and flow rises, but the higher flow raises frictional pressure drop until a new steady flow is reached. Temperature in a heated vessel behaves the same way, as does pressure in a system with a natural outlet. The hallmark is self-limiting behavior; a fixed output produces a fixed final value.
An integrating process has no such balancing tendency, because it is governed by accumulation. Tank level is the textbook case: level is the running total of everything that has flowed in minus everything that has flowed out. If inflow exceeds outflow by even a little, level rises steadily and keeps rising for as long as the imbalance persists; it never finds a resting point on its own, because nothing about the rising level acts to close the gap between in and out. Pressure in a sealed vessel behaves the same way, integrating the net mass flow. A fixed output does not give a fixed level; it gives a constant rate of change of level.
This difference is not cosmetic. In a self-regulating process, the controller's job is to find the output that produces the desired steady value. In an integrating process, any steady output other than the exact balance point drives the variable off without limit, so the controller must continuously hunt for the precise output that makes inflow equal outflow. The very definition of what the controller is trying to do changes with the process type.
The cleanest way to tell the two apart is a bump test with the loop in manual. Step the controller output and watch the measurement. A self-regulating process rises (or falls) and then levels off at a new steady value, tracing the familiar S-shaped reaction curve that flattens out. An integrating process does something visually striking: instead of leveling off, the measurement keeps ramping in a straight line, its slope proportional to the size of the output step. If the trend after a step is a curve that settles, the process is self-regulating; if it is a ramp that will not settle, the process is integrating. That ramp is the signature.
The tuning consequences are significant. Most classic tuning rules and process models assume a self-regulating process with a definable steady-state gain, the ratio of final measurement change to output change. An integrating process has no steady-state gain in that sense, because the measurement never reaches a final value; it is characterized instead by an integrating rate. Applying self-regulating tuning math to an integrating loop, or trying to read a steady-state gain off a ramp that never flattens, produces nonsense parameters and a badly tuned loop. Integrating processes need their own tuning rules built around the integrating rate and typically call for gentler, more heavily damped settings because they are prone to slow, rolling oscillations.
Misclassifying the process type is one of the most common tuning failures precisely because level loops are everywhere and are integrating, while the intuition most people carry is built on self-regulating flow and temperature loops. An engineer who treats a level loop like a flow loop will often end up with a loop that cycles slowly and never quite settles, chasing the problem in the tuning constants when the real error was upstream, in the classification. Naming the process type correctly is the first step, and skipping it undermines everything that follows.
Process type is not only a tuning concern; it also shapes how a loop should be monitored and alarmed, which is squarely a SCADA question. An integrating variable like tank level or vessel pressure can march steadily toward a limit while looking calm on a snapshot, because a small steady imbalance produces a slow, relentless ramp rather than an obvious swing. Monitoring logic that expects a variable to settle can be lulled by that quiet ramp, so understanding that a loop is integrating informs how its trends should be watched and how far ahead its alarms should look.
In a cloud SCADA platform such as Merobix, the historized trends for a loop reveal its type at a glance and let analysis tools apply the right treatment. A performance-monitoring layer that knows level and closed-vessel pressure loops are integrating can assess them against integrating benchmarks rather than misjudging them by self-regulating standards. For assets that run unattended, remote tank farms, wellpad separators, water reservoirs, that correct framing keeps a slow-ramping integrating variable from sliding toward a high or low limit unnoticed between the infrequent human check-ins.
The distinction spans every industry these platforms serve. Oil and gas separators and surge tanks, water and wastewater basins, boiler drums in power plants, and buffer vessels in manufacturing are all integrating level or pressure loops, while their flow and temperature loops are self-regulating. Recognizing which is which, in both the tuning and the monitoring, is foundational, and a monitoring system that respects the difference reads each loop's behavior correctly instead of forcing every loop into the same mold.
A self-regulating process settles to a new steady state on its own after a step in the controller output, because its own response opposes further change; flow and temperature are typical examples. An integrating process ramps continuously and never settles by itself, because it accumulates a net imbalance; tank level and closed-vessel pressure are typical examples. A fixed output gives a fixed value in the first case and a constant rate of change in the second.
Step the controller output with the loop in manual and watch the measurement. If it rises or falls and then levels off at a new steady value, the process is self-regulating. If it keeps ramping in a straight line without settling, its slope set by the step size, the process is integrating. The ramp that will not flatten is the signature of an integrating process.
Level is an integrating process, so it has no steady-state gain and never settles at a fixed output the way a self-regulating flow or temperature loop does. Standard tuning rules that assume a settling response and a defined steady-state gain give wrong parameters when applied to it. Integrating loops need rules built around the integrating rate and usually gentler, more damped settings to avoid slow rolling oscillations.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.