How to Troubleshoot Ratio Control: Causes, Tests and Solutions

Figure V01. Ratio-control troubleshooting begins by checking the complete path from measured flows and setpoint arithmetic to the final control element. Original self-drawn illustration

Text alternative for this figure

Engineer with an open laptop beside two flow-controlled pipelines with transmitters and valves merging into a process mixer at sunrise.

Educational engineering guide · Updated 2026-10-11 · Worked-example data are illustrative unless labelled SOURCE.

Quick answer

To troubleshoot ratio control, start by confirming the intended relationship: the controlled-flow setpoint should normally be calculated as the ratio multiplied by the measured wild flow. In symbols, SP_controlled=R× F_wild, where R=F_controlled/F_wild. Then check the system layer by layer: safety and operating state, modes and tracking, setpoint arithmetic and scaling, flow measurement, valve capacity and movement, process dynamics, ratio trim, and analyzer behavior.

Do not begin by retuning the controller. First compare the expected controlled-flow setpoint with the value actually generated by the ratio station. If the setpoint is wrong, investigate ratio direction, engineering-unit conversion, span scaling, square-root processing, density basis, bias, trim, selectors, and low-flow handling. If the setpoint is correct but the controlled flow does not follow it, investigate controller mode, output limits, valve saturation, stem response, stiction, supply pressure, bypass leakage, and dynamic mismatch.

Key takeaways

Define the ratio explicitly as controlled flow divided by wild flow before testing anything.
In the usual architecture, multiply wild flow by the ratio; do not divide a flow by another flow to generate the controlled-flow setpoint.
Verify engineering units, measurement basis, ranges, and span conversion before changing tuning.
A differential-pressure flow signal needs exactly one square-root operation somewhere in its configured signal path.
Ratio calculations become unreliable near zero wild flow because noise and division sensitivity become dominant.
If the calculated setpoint is correct but the controlled flow is wrong, focus on modes, limits, measurement, and the final control element.
Valve saturation, stiction, installed characteristic, and bypass leakage can resemble controller-tuning problems.
Dynamic mismatch creates temporary composition or blending errors even when both steady-state flow readings are correct.
Ratio trim should be slow, bounded, and applied only after the inner flow loops work correctly.
Combustion cross-limiting coordinates air and fuel demand, but it does not replace burner-management protection.

How ratio control works: wild vs controlled flow

Do this first: Write down the ratio definition, its units, the wild-flow variable, the controlled-flow variable, and the expected setpoint equation.

A ratio-control system maintains a selected relationship between two measured quantities. The most common process example uses one flow that changes independently and another flow that must follow it. The independently changing flow is often called the wild flow, master flow, or uncontrolled flow. The manipulated flow is the controlled flow, secondary flow, or follower flow.

For the convention used throughout this guide:

R = F_controlled / F_wild

SP_controlled = R × F_wild

If the desired controlled flow is 5% of the wild flow, then R=0.05. At a wild flow of 400 m³/h, the required controlled-flow setpoint is 20 m³/h:

SP_controlled = 0.05 × 400 m³/h = 20 m³/h

This multiplication architecture is described in educational and practitioner explanations from LibreTexts/11%3A_Control_Architectures/11.04%3A_Ratio_control-_What_is_it_When_useful_When_not_Common_usage), Control.com, and ControlGuru SOURCE. It is usually implemented as a controlled-flow loop in cascade with a ratio station. The ratio station generates the remote setpoint, while the secondary flow controller manipulates the valve, speed, or other final element.

A typical signal path is:

The wild-flow transmitter measures the wild stream.
The control system scales the measurement into engineering units or a normalized linear-flow signal.
The ratio station multiplies that flow by the selected ratio.
Optional bias, trim, limits, selectors, and tracking logic modify or constrain the result.
The resulting value becomes the remote setpoint of the controlled-flow controller.
The controlled-flow controller moves its final element until its measured flow follows the generated setpoint.

Figure V02 shows the relationship between the wild stream, ratio station, controlled-flow controller, flow transmitter, and final control element.

Figure V02. The measured wild flow establishes the follower-flow setpoint through the configured ratio relationship (conceptual picture). Conceptual picture — no data

Text alternative for this figure

A broad coral ribbon (the wild flow) passes a round measurement element, and a narrower sage ribbon (the follower flow) passes a valve. A dotted arrow runs from the measurement element to the follower’s control element. Both ribbons enter a central mixing vessel. It is a conceptual picture and carries no data.

Ratio station and ratio definition

The ratio station may be a dedicated function block, part of a distributed control system strategy, or logic assembled from multiply, bias, limit, and tracking blocks. Its displayed ratio might be an engineering ratio such as kilograms per kilogram, a percentage, or a normalized gain. Never infer the internal definition from the display label alone.

Project documentation may identify ratio functions with tags such as FFY or FY, but the project tagging standard governs the actual tag. The guides to instrument tags and function symbols such as multipliers and selectors can help with drawing interpretation, but the control narrative and configured logic remain decisive.

Check whether the configured ratio means:

controlled flow divided by wild flow;
wild flow divided by controlled flow;
percentage controlled flow relative to wild flow;
normalized output gain;
mass ratio despite volumetric displays;
ratio after density, temperature, pressure, or composition correction; or
a ratio of previously calculated ratios.

An inverse ratio is legitimate when deliberately defined. For example, a project may define R_inverse=F_wild/F_controlled. In that case, the implementation must invert it before calculating the controlled setpoint, or use equivalent logic. The problem is not inverse notation itself; the problem is an undocumented or inconsistently applied convention.

A ratio of ratios can also be valid in specialized blending or composition strategies. It requires careful unit analysis because each intermediate quantity may use a different basis. Document the complete equation rather than relying on a short label such as “ratio trim.”

Why direct division is usually the wrong control architecture

A calculated ratio display can be useful:

Ratio_PV = F_controlled / F_wild

However, using this divided value as the primary controlled variable of a proportional-integral-derivative controller introduces a serious low-flow problem. Its gain with respect to controlled flow is:

d(Ratio_PV) / dF_controlled = 1 / F_wild

As wild flow approaches zero, the calculated sensitivity approaches infinity. Measurement noise, offsets, quantization, and small leakage flows can therefore produce very large apparent ratio changes.

At F_wild=400 m³/h and F_controlled=20 m³/h, a ratio display correctly reports 0.05. That does not mean a controller should manipulate the valve by controlling the divided value directly. The multiplier architecture instead generates a stable controlled-flow setpoint of 20 m³/h and lets a conventional flow loop perform the regulation.

Figure V03 contrasts the preferred multiplier arrangement with direct ratio-PV control for R=0.05, F_wild=400 m³/h, and SP_controlled=20 m³/h. It also shows why the division gain becomes extreme near zero wild flow.

Figure V03. Generate the follower setpoint by multiplication; reserve division-based ratio calculations for display. Original self-drawn diagram

Text alternative for this figure

Left, correct: measured wild flow F_wild enters a multiplier with R = 0.05, giving SP = R × F_wild. With F_wild = 400 m³/h the setpoint is 20 m³/h, feeding the follower flow PID and control valve. Right, risky: F_controlled divided by F_wild forms a ratio PV for a ratio PID; the gain with respect to controlled flow is 1/F_wild, which tends to infinity as wild flow approaches zero. Division is acceptable for display only, with a low-flow guard.

Ratio on flow versus ratio on another measured value

Flow is the usual ratio-control variable because it responds promptly and is directly related to material throughput. The required relationship may nevertheless depend on mass, energy, concentration, moisture, heating value, or composition rather than raw volumetric flow.

An analyzer can adjust a base ratio, but using a slow analyzer as if it were a fast flow measurement can create poor dynamics. A practical architecture often has:

fast inner flow-control loops;
a base feedforward ratio derived from throughput;
slow, bounded analyzer trim; and
limits that prevent the trim from requesting an unsafe or impossible flow.

The distinction is important. Flow ratio controls the immediate material-rate relationship. Analyzer trim corrects slower deviations in the result. If the controlled flow cannot follow its setpoint, fixing analyzer tuning will not solve the inner problem.

Ratio control versus feedforward control

Ratio control is a structured form of feedforward action because it measures a disturbance—the wild flow—and changes another setpoint in anticipation of its effect. General feedforward control can use more complex process models, multiple measured disturbances, nonlinear calculations, or dynamic compensation. Ratio control uses a defined proportional relationship, optionally supplemented by bias, trim, and lead/lag.

Ratio control is not the right tool when the required manipulated quantity is unrelated to the measured wild flow, when the wild-flow measurement is unreliable, when unmeasured disturbances dominate, or when the process relationship is strongly nonlinear and cannot be represented safely by scheduled ratios or an approved model.

Readers who need drawing fundamentals can review control loops on a P&ID and the broader instrument-symbol and identification conventions.

Triage: the first ten minutes

Do this first: Stabilize the process, confirm the operating envelope and protection status, and capture the current values before anyone changes modes, ratios, limits, or tuning.

Fast ratio control troubleshooting depends on testing in the right order. Start with conditions that can make all later observations misleading: unsafe operation, manual modes, disabled remote setpoints, stale tracking values, bad scaling, output saturation, and invalid low-flow calculations.

Figure V04 presents the recommended sequence: safety, mode and tracking, setpoint arithmetic, scaling, measurement, saturation, valve movement, dynamics, and trim.

Figure V04. Troubleshoot in sequence so configuration and measurement faults are excluded before tuning or trim changes. Original self-drawn diagram

Text alternative for this figure

Safety: safe to investigate, loop type known, approvals in place. If not safe, stop and escalate.
Mode and tracking: controlled loop in auto or remote, ratio station and selectors tracking.
Setpoint arithmetic and scaling: recalculate SP = R × F_wild by hand; check direction, units, spans, bias, trim.
Measurement: ranges, square root once, low-flow cutoff, density, temperature and pressure basis, noise.
Valve saturation: output at a limit, capacity, supply pressure, bypass leakage.
Stem response and stiction.
Dynamics: step-test each flow path.
Trim and analyzer: authority, delay, filtering; tune after the inner loops.
At each step: fault found means fix here and retest.

Use this initial sequence:

Confirm the process is in an approved state and identify all applicable alarms, interlocks, overrides, and safety functions.
Determine whether the controlled-flow loop is in manual, local, automatic, cascade, or remote-setpoint mode.
Record the wild flow, selected ratio, bias, trim, calculated setpoint, controlled-flow process value, controller output, and valve stem feedback.
Calculate the expected setpoint independently.
Compare the expected value with the ratio station’s generated value.
If those values differ, investigate arithmetic, direction, scaling, units, tracking, limits, and selectors.
If the generated setpoint is correct, compare it with controlled-flow PV.
If PV cannot reach setpoint, inspect mode, saturation, valve movement, measurement, capacity, and process constraints.
If steady-state tracking is good but the blend deviates during changes, investigate dynamics and delays.
Only after these checks should tuning or analyzer trim be changed.

Symptom-to-test triage table

Symptom	Likely layer	First test	Typical corrective direction
Generated setpoint has the wrong magnitude	Ratio arithmetic or scaling	Calculate `R× F_wild` independently	Correct ratio direction, units, spans, or normalization
Setpoint changes in the wrong direction	Ratio direction or action	Increase a simulated or approved test input slightly	Correct inversion, sign, or selector logic
Flow PV is below a correct setpoint	Mode, output limit, valve, or capacity	Check mode, output, stem feedback, and saturation	Restore cascade, remove approved constraint, or repair capacity issue
Flow PV is above a correct setpoint	Valve leakage, minimum output, or bad measurement	Compare command, stem, isolated-flow evidence, and transmitter zero	Repair leakage, zero error, or output-limit configuration
Ratio is erratic near shutdown	Low-flow measurement and division	Trend both raw flows and calculation validity	Apply approved cutoff, validity logic, or display suppression
Ratio is wrong at all loads by a constant factor	Scaling or basis	Compare raw input, scaled PV, spans, and units	Correct scaling or basis conversion
Error grows with load	Valve capacity, pressure loss, or range limit	Trend output and stem position against throughput	Correct final-element sizing or operating constraint
Error appears only during ramps	Dynamic mismatch	Compare response delays and time constants separately	Apply tested dynamic compensation or slower demand shaping
Slow oscillation remains after flows settle	Trim or analyzer	Place trim in an approved fixed state and observe inner loops	Retune only after verifying analyzer delay and calibration
Auto transfer causes a bump	Tracking or initialization	Compare applied setpoint with stored ratio and bias	Track inactive blocks to the applied value

Twenty-case screening output

The following table is an illustrative screening dataset. Ratio error is calculated as:

Ratio error (%) = 100 × (actual − set) / set

Case	Symptom	Ratio error	Script flags	Suspected layer
C01	normal operation	0.0%	none	none
C02	controlled flow far below expected	-87.5%	SCALING_SUSPECT	configuration: forgot span ratio
C03	controlled flow double expected	+100.0%	SCALING_SUSPECT	configuration: ratio direction reversed
C04	controlled flow cannot reach setpoint	-28.6%	SATURATION	final element: valve saturated high
C05	controlled flow remains high	+50.0%	SATURATION, SCALING_SUSPECT	final element: valve saturated low or bypass leaking
C06	ratio noisy near shutdown	+50.0%	LOW_FLOW_RISK, SCALING_SUSPECT	measurement: DP square root near zero
C07	ratio undefined at zero wild flow	NA	SATURATION, LOW_FLOW_RISK	design: division by zero
C08	setpoint not followed	-20.0%	MODE_TRACKING	mode: controlled loop in manual
C09	bump on auto transfer	-60.0%	MODE_TRACKING, SCALING_SUSPECT	initialisation: ratio station not tracking
C10	slow periodic cycling	-4.0%	none	final element: 2 percent stiction
C11	ratio droops at high load	-20.0%	SATURATION	final element: insufficient valve rangeability
C12	ratio high only when hot	+10.0%	none	measurement: density uncompensated
C13	transient lean then recovers	-10.0%	none	dynamics: controlled path too slow
C14	steady offset despite trim	+10.0%	none	trim: authority exhausted
C15	trim oscillates slowly	+12.0%	none	analyzer: delay ignored
C16	ratio PV implausibly high	+100.0%	LOW_FLOW_RISK, SCALING_SUSPECT	design: division-based ratio PV at low flow
C17	controlled flow fixed after cascade break	-20.0%	MODE_TRACKING	mode: remote setpoint disabled
C18	flow reads low across range	-50.0%	SCALING_SUSPECT	measurement: square root missing
C19	flow reads high at midrange	+40.0%	none	measurement: square root applied twice
C20	normal with positive trim	+10.0%	none	trim: at positive limit

The companion script and CSV are included in the package. They are screening aids, not diagnostic proof. A single snapshot cannot identify density changes, dynamic mismatch, analyzer delay, or stiction reliably. In this dataset, the script misses C10, C12, C13, C14, C15, and C19. It also over-flags some cases: in C06, “SCALING_SUSPECT” is generated by the numerical pattern, but the intended problem is low-flow noise around a differential-pressure square-root calculation.

A flag should therefore trigger a targeted test, not an immediate configuration change. For example, a saturation flag should be checked against controller output, configured limits, stem feedback, available pressure drop, and process demand. A scaling flag should be checked against raw counts, input range, engineering range, square-root location, and the configured ratio equation.

Preserve evidence before intervention

Capture enough trend history to show:

the last stable period;
the start of the deviation;
any throughput or mode change;
wild-flow PV;
controlled-flow setpoint and PV;
raw transmitter signals if available;
controller output and selected output;
stem position;
active limits and selectors;
ratio, bias, and trim values;
analyzer PV, quality, and calibration status; and
alarms, interlocks, permissives, and operator actions.

Do not reset blocks merely to remove a symptom. A reset may erase the state that distinguishes stale tracking from windup, or valve saturation from a transmitter problem.

Design and scaling errors

Do this first: Recalculate the ratio station from documented engineering ranges and compare it with the live block configuration.

Scaling errors often produce clean, repeatable, but completely wrong behavior. Because the loop may remain stable and responsive, the problem can be mistaken for an incorrect process recipe, valve problem, or transmitter bias.

Engineering units versus normalized signals

If the ratio calculation uses engineering-unit flows, the equation can be direct:

SP_controlled = R × F_wild

If it operates on normalized percentages, the span relationship must be included. For zero-based ranges:

K = R × (wild-flow span / controlled-flow span)

Controlled SP (%) = K × Wild PV (%)

This conversion is easy to miss because a ratio of 0.05 appears intuitively suitable as a percentage-domain gain. It is suitable only when the two normalized spans are equal or when another block already accounts for the span ratio.

For non-zero-based ranges, use full range conversions rather than assuming percentage equals flow divided by upper range. Also identify whether live zero, suppressed zero, bidirectional range, square-root extraction, or custom characterization is present.

Worked example: forgotten span ratio

Consider an additive-to-wild-stream volumetric ratio ILLUSTRATIVE:

Desired ratio: 0.05 ILLUSTRATIVE
Wild-flow range: 0–400 m³/h ILLUSTRATIVE
Additive-flow range: 0–50 m³/h ILLUSTRATIVE

The correct normalized gain is:

K = 0.05 × (400 / 50)

K = 0.4 CALCULATED

At a wild flow of 400 m³/h:

SP_additive = 0.05 × 400 m³/h

SP_additive = 20 m³/h CALCULATED

Relative to the additive span:

Additive SP = (20 / 50) × 100%

Additive SP = 40% of span CALCULATED

If the span ratio is forgotten and the configured normalized gain is incorrectly set to 0.05, then a 100% wild-flow signal generates only a 5% additive-flow setpoint:

Wrong additive flow = 0.05 × 50 m³/h

Wrong additive flow = 2.5 m³/h CALCULATED

The actual ratio becomes:

Actual ratio = 2.5 / 400

Actual ratio = 0.00625 CALCULATED

The error relative to the intended ratio is:

Error = 100 × (0.00625 − 0.05) / 0.05

Error = −87.5% CALCULATED

Wild flow (m³/h) ILLUSTRATIVE	Correct additive flow (m³/h) CALCULATED	Correct additive span (%) CALCULATED	Wrong-K additive flow (m³/h) CALCULATED
0	0	0	0
100	5	10	0.625
200	10	20	1.25
300	15	30	1.875
400	20	40	2.5

The constant factor across the operating range is a strong scaling clue. Confirm it at more than one point because valve saturation or a nonlinear meter can conceal the pattern.

Bias changes the instantaneous ratio

A ratio station may include an additive bias:

SP_controlled = R × F_wild + B

For an illustrative base ratio of 0.05 and bias of 2 m³/h ILLUSTRATIVE:

SP_controlled = 0.05 × F_wild + 2 m³/h

Wild flow (m³/h) ILLUSTRATIVE	Controlled-flow SP (m³/h) CALCULATED	Instantaneous ratio CALCULATED
0	2	Not meaningful at zero wild flow
100	7	0.07
200	12	0.06
300	17	0.0567
400	22	0.055

A bias therefore changes the instantaneous ratio with throughput. It may be appropriate for a documented physical requirement, such as a fixed base contribution, but it should not be described as maintaining one constant ratio.

At zero wild flow, the equation requests 2 m³/h. Whether that is allowed depends on process requirements, shutdown logic, minimum-flow logic, and safeguards. The calculation alone does not establish that such operation is safe or intended.

Ratio trim authority

Trim changes the base ratio within defined limits. For a base ratio of 0.05 and trim authority of ±10% ILLUSTRATIVE:

Minimum ratio = 0.05 × (1 − 0.10) = 0.045 CALCULATED

Maximum ratio = 0.05 × (1 + 0.10) = 0.055 CALCULATED

At 200 m³/h wild flow:

Minimum controlled SP = 0.045 × 200 = 9 m³/h CALCULATED

Maximum controlled SP = 0.055 × 200 = 11 m³/h CALCULATED

This is multiplicative trim. An additive trim applied directly to the ratio or flow setpoint produces different limits. Record the exact block equation and whether the displayed “10%” means percent of base ratio, percentage points of ratio, percent of flow span, or percent of current setpoint.

Direction reversal and basis errors

A reversed definition can produce large errors without any failed equipment. If the process requires controlled/wild = 0.05 but the configured station interprets its entry as wild/controlled, the intended inverse would be 20. Entering 0.05 into an inverse-ratio block is not a small tuning mistake; it defines a different relationship.

Both streams must also be expressed on a compatible basis:

mass with mass;
actual volume with actual volume under compatible conditions;
standard volume with standard volume at stated reference conditions;
molar flow with molar flow; or
a documented conversion connecting different bases.

Watch for mismatches such as kg/h versus t/h, L/min versus m³/h, actual m³/h versus standard m³/h, or wet basis versus dry basis. Unit labels on operator graphics do not prove that internal signals use those units.

Scaling verification procedure

Record transmitter lower and upper range values.
Record input-card raw range and configured engineering range.
Identify where square-root extraction occurs.
Record all normalization, characterization, density, temperature, and pressure blocks.
Trace the wild signal to the ratio station.
Recalculate the ratio station output at 0%, 25%, 50%, 75%, and 100% input.
Trace the resulting value through bias, trim, limits, selectors, and remote-setpoint scaling.
Verify the controlled-flow controller receives the expected engineering value.
Confirm the operator display reports the same ratio convention.
Save the approved configuration and test results.

Figure V05 shows how the same physical ratio passes through engineering-unit and percentage-domain implementations, and where the span factor belongs.

Figure V05. A raw ratio of 0.05 becomes a normalised gain of 0.4 when the two flow ranges are 400 and 50 (illustrative values). Illustrative / calculated

Text alternative for this figure

R = 0.05 additive to wild by volume; wild span 0–400 m³/h, additive span 0–50 m³/h; K = 0.05 × 400/50 = 0.4. Correct additive setpoint at wild flow 0, 100, 200, 300, 400 m³/h: 0, 5, 10, 15, 20 m³/h (0, 10, 20, 30, 40 % of span). With the wrong K of 0.05 the additive is 0, 0.625, 1.25, 1.875, 2.5 m³/h; at 400 m³/h the actual ratio is 0.00625, an error of −87.5 %.

Flow measurement problems

Do this first: Compare the raw transmitter signal, scaled linear-flow PV, and an independent expected flow before touching ratio or controller tuning.

A ratio calculation can only be as accurate and responsive as its input measurements. Two individually acceptable meters may still produce a poor ratio if they use different physical bases, have mismatched response times, operate below useful range, or apply inconsistent compensation.

Range and usable operating region

A transmitter configured for a very wide range may provide inadequate resolution or signal quality at the actual operating point. Conversely, a narrow range may overrange during high production. Check:

sensor and transmitter range;
configured range;
normal, minimum, and maximum operating flows;
transmitter damping;
control-system filtering;
low-flow cutoff;
totalizer behavior;
alarm and bad-quality handling; and
whether the ratio station uses the same PV displayed to the operator.

Do not treat a catalog turndown statement as proof of installed performance. Installation, impulse lines, upstream disturbances, density assumptions, zero stability, and process conditions matter.

Differential-pressure square-root processing

For an idealized head-type differential-pressure flow measurement:

Q / Qmax = sqrt(dP / dPmax)

This relationship is described in Control.com’s differential-pressure flow explanation SOURCE. The signal path must perform the required square root exactly once. If the transmitter outputs raw differential pressure, another configured element must linearize it. If the transmitter already outputs linearized flow, a second square root is wrong.

Missing square root corresponds to case C18 in the screening table. Applying the square root twice corresponds to C19.

Differential pressure (% of span)	Flow (% of maximum) CALCULATED
0	0
1	10
4	20
9	30
16	40
25	50
36	60
49	70
64	80
81	90
100	100

A missing square root makes indicated flow too low below full scale. For example, 25% differential pressure represents 50% flow, not 25% flow. A duplicate square root makes the displayed value too high through much of the range.

Why low-flow noise becomes severe

Consider fixed differential-pressure uncertainty of ±0.5% of DP span ILLUSTRATIVE. With normalized flow x expressed in percent, local linearization gives a sensitivity equivalent to:

dFlow / dDP = 50 / x

The resulting illustrative local errors are:

Actual flow (% of maximum) ILLUSTRATIVE	Local flow error (percentage points) CALCULATED	Relative flow error (%) CALCULATED
100	0.25	0.25
50	0.50	1.00
25	1.00	4.00
10	2.50	25.00
5	5.00	100.00

This is a local linearization, not an exact uncertainty model. It breaks down near zero, where negative noisy differential-pressure values, transmitter clamping, cutoff logic, and nonlinear extraction dominate. The table demonstrates why a ratio display may become unusable before the transmitter appears obviously failed.

Filtering can reduce visible noise, but it adds lag. Excessive filtering on one flow and not the other creates a transient ratio error during throughput changes. Correct the measurement source and coordinate response times before applying heavy filtering.

Low-flow cutoff

A low-flow cutoff suppresses or modifies output below a configured threshold. It can prevent noise from appearing as flow, but it can also conceal genuine leakage or small process flow. The threshold, hysteresis, alarm behavior, totalization behavior, and restart transition must therefore be documented.

An illustrative cutoff at 5% flow corresponds to 0.25% differential pressure ILLUSTRATIVE. This is not a recommended default.

A Yokogawa application note describes a product-specific low-flow cutoff programmable from 0–20% of full scale, with an example using 5% and nominal 1% hysteresis SOURCE. Those figures describe that product example; they are not universal settings.

Test both sides of a cutoff:

decreasing flow through the threshold;
increasing flow back through the threshold;
exact zero-flow condition;
transmitter bad quality;
shutdown and restart;
ratio-station tracking while cut off; and
valve response when the signal becomes valid again.

Division sensitivity at low wild flow

For a displayed ratio calculated by dividing controlled-flow percentage points by wild-flow percentage, the gain per percentage point of controlled flow is:

Wild flow (%) ILLUSTRATIVE	Ratio-display gain per percentage point of controlled flow CALCULATED
100	0.010
50	0.020
10	0.100
5	0.200
1	1.000

At 1% wild flow, a one-percentage-point controlled-flow change moves the calculated ratio by 1.000. This mathematical sensitivity explains implausibly high ratio displays without proving either transmitter has failed.

Use validity logic for displays and diagnostics. The control strategy should normally generate the follower setpoint by multiplication and define approved behavior when wild flow is too low for a valid ratio.

Mass versus volume

A volumetric ratio does not automatically equal a mass ratio. Consider ILLUSTRATIVE:

wild-stream density: 1000 kg/m³ ILLUSTRATIVE;
additive density: 800 kg/m³ ILLUSTRATIVE; and
volumetric additive-to-wild ratio: 0.05 ILLUSTRATIVE.

The corresponding mass ratio is:

Mass ratio = 0.05 × 800 / 1000

Mass ratio = 0.0400 CALCULATED

If additive density drops by 3%:

New additive density = 800 × 0.97 = 776 kg/m³ CALCULATED

New mass ratio = 0.05 × 776 / 1000 = 0.0388 CALCULATED

Mass-ratio change = −3.0% CALCULATED

At 100 m³/h wild flow:

Wild mass flow = 100 m³/h × 1000 kg/m³ = 100,000 kg/h CALCULATED

The additive volumetric flow is:

Additive volume flow = 0.05 × 100 m³/h = 5 m³/h CALCULATED

Before the density change:

Additive mass flow = 5 m³/h × 800 kg/m³ = 4,000 kg/h CALCULATED

After the density change:

Additive mass flow = 5 m³/h × 776 kg/m³ = 3,880 kg/h CALCULATED

The volumetric ratio remains 0.05 while the delivered additive mass changes. Whether compensation is required depends on what physical quantity the process specification actually controls.

Gas-flow comparisons require particular care. Standard or normal volumetric flow must state its reference temperature and absolute pressure. Actual volumetric flow depends on operating pressure, temperature, composition, and compressibility. Do not compare two gas-volume signals merely because both displays use m³/h.

Impulse lines, installation, and signal conditioning

For differential-pressure installations, inspect approved maintenance records and indications related to:

plugged or partially blocked impulse lines;
condensation or vapor pockets;
leaking manifolds;
unequal liquid heads;
heat tracing;
transmitter zero shift;
reversed connections;
pulsation;
wet-leg condition; and
process-fluid density changes.

For other meter technologies, check their applicable installation conditions, signal quality, diagnostics, grounding, straight-run requirements, coating, entrained phases, and compensation inputs.

Figure V06 maps common measurement symptoms to range, square-root, cutoff, density, impulse-line, and filtering checks.

Figure V06. Fixed DP noise produces a progressively larger relative flow error as flow falls toward the low end of range (illustrative noise). Illustrative noise; calculated

Text alternative for this figure

Left: DP percent 0, 1, 4, 9, 16, 25, 36, 49, 64, 81, 100 gives flow percent 0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100. Right: fixed ±0.5 % DP noise gives relative flow errors of 0.25, 1.00, 4.00, 25.00 and 100.00 % at 100, 50, 25, 10 and 5 % flow. An illustrative cutoff of 5 % flow equals 0.25 % DP and is not a recommended default.

Valve problems

Do this first: Trend controlled-flow setpoint, flow PV, controller output, selected output, and actual stem position on the same time axis.

When the generated setpoint is correct but measured flow cannot follow it, the final control element becomes a primary suspect. That does not mean the valve itself is always faulty. Output limits, low supply pressure, positioner configuration, pressure-drop changes, downstream restrictions, pump limitations, and bypass leakage can produce similar symptoms.

Saturation and capacity

A controller is saturated when its demanded output is held at an upper or lower limit. Determine which limit is active and where it is applied:

controller output limit;
selected override output;
valve position limit;
positioner travel limit;
drive-speed limit;
permissive constraint;
split-range boundary; or
physical process capacity.

If controller output is high, stem feedback is high, and flow remains below setpoint, possible causes include insufficient valve capacity, inadequate upstream pressure, downstream restriction, wrong valve characteristic, plugged equipment, or a biased flow measurement.

If output is low and measured flow remains high, investigate bypass leakage, valve-seat leakage, fail-position behavior, incorrect action, minimum-flow paths, another active source, and transmitter zero error.

Stiction and stick-slip

Static friction can prevent the stem from responding to small command changes. Controller integral action continues accumulating until the command change exceeds the sticking threshold. The stem then moves suddenly, the process overshoots, and the cycle repeats.

Useful evidence includes:

controller output ramps while stem feedback remains nearly constant;
sudden stem jumps;
a repeating output-PV pattern;
different response when approaching a position from opposite directions; and
oscillation that changes with valve position or pressure drop.

A demonstration model uses a 2% span stick threshold, process time constant of 20 s, proportional-integral settings K_p=1.2 and K_i=0.8 per second ILLUSTRATIVE. After its transient, the modeled PV cycles approximately from 49.53% to 50.36%, with a half-amplitude of about 0.42 percentage points and a period of about 42 s CALCULATED.

These values describe demonstration behavior, not an acceptance criterion. There is no universal stiction limit. The model’s 2% threshold must not be applied as a plant standard.

Figure V08 shows the illustrative stick-slip cycle as the controller output winds across the sticking band and the stem moves in discrete jumps.

Figure V08. Stiction is indicated when the command keeps moving while the stem position sticks and then releases abruptly (illustrative model). Illustrative model — not an acceptance limit

Text alternative for this figure

Three aligned trends over 100–200 s. The controller command moves smoothly; the valve position stays fixed and then jumps by about 2 percentage points (the illustrative stick threshold) when the command has moved far enough; the process variable cycles between about 49.5 and 50.4 % with a period of about 42 s.

Installed characteristic and controllable range

The inherent valve characteristic is measured under defined test conditions. The installed relationship between position and flow also depends on system pressure losses and operating conditions. A nominal equal-percentage, linear, or quick-opening characteristic does not by itself establish the installed controllable range.

ISA-75.11 addresses definitions and tests associated with inherent flow characteristic and rangeability. The ISA-75.11 committee page is relevant for scope SOURCE, but this guide does not reproduce the paid standard or assign numerical acceptance limits.

Treat rangeability as the installed controllable range needed by the application, not merely a catalog figure. Test whether small position changes produce repeatable flow changes at low, normal, and high load.

Practical valve checks

With approval and within the operating envelope:

Hold other major disturbances as steady as practical.
Record the current mode, output, stem, pressure conditions, and flow.
Apply a small authorized output bump or setpoint step.
Confirm that selected output changes as expected.
Confirm stem feedback begins moving in the correct direction.
Compare command movement with stem movement.
Compare stem movement with flow response.
Repeat from the opposite direction if the procedure allows.
Restore the approved mode and setpoint.
Save the trend and record any hysteresis, delay, saturation, or nonmovement.

A command that moves while the stem does not points toward positioner, actuator, friction, supply, or linkage problems. A stem that moves while flow barely changes points toward process pressure, installed gain, blockage, measurement, or valve sizing. Flow that persists at a closed command points toward leakage, bypass paths, fail action, another source, or measurement zero.

Dynamic mismatch and lead/lag

Do this first: Test and identify the wild and controlled paths separately rather than inferring both from a single ratio trend.

Ratio arithmetic acts immediately, but physical processes do not. The wild flow may reach the mixing point sooner than the controlled flow because the streams have different valves, pipes, equipment volumes, transport delays, sensors, filters, and inner-loop responses.

A ratio can therefore be correct at steady state while the mixed composition deviates during ramps. Retuning the ratio multiplier cannot remove a transport delay.

Separate step tests

An approved step test should identify, for each path:

steady-state gain;
apparent dead time;
dominant time constant;
direction;
valve saturation;
measurement filtering;
nonlinear behavior; and
dependence on operating point.

Use small steps that remain within the approved envelope. Change one path at a time where the process and procedure allow it. Record the command, setpoint, PV, output, stem position, and downstream response.

Do not assume the controlled path is slow merely because the downstream analyzer responds late. The analyzer sample system may add tubing delay, sample conditioning time, cycle time, filtering, and validation delay.

Worked example: mismatched paths

Consider an illustrative normalized blending model ILLUSTRATIVE:

wild-path time constant: 30 s ILLUSTRATIVE;
wild-path dead time: 5 s ILLUSTRATIVE;
controlled-path time constant: 90 s ILLUSTRATIVE;
controlled-path dead time: 15 s ILLUSTRATIVE;
both steady-state flows change from 1.0 to 1.5 ILLUSTRATIVE.

Composition is defined as:

C = c / (w + c)

Composition error in percentage points is:

Error = 100 × (C − 0.5)

Without dynamic compensation, the calculated peak composition error is 4.5457 percentage points CALCULATED. The error settles within ±0.1 percentage point at 412.0 s CALCULATED.

A lead/lag compensator is then applied:

H(s) = (90s + 1) / (30s + 1)

This example has a lead of 90 s and a lag of 30 s ILLUSTRATIVE. With that compensation, the calculated peak error is 3.3308 percentage points, and the error settles within ±0.1 percentage point at 109.5 s CALCULATED.

The remaining dead-time difference is:

Dead-time difference = 15 s − 5 s = 10 s CALCULATED

That 10 s difference remains uncompensated. A lead/lag block reshapes a measured signal, but it cannot predict an unmeasured future change.

Figure V07 compares the uncompensated and compensated normalized responses, including the remaining dead-time mismatch.

Figure V07. Illustrative lead/lag compensation reduces the peak error and settling time, although a 10 s dead-time mismatch remains. Illustrative model — calculated

Text alternative for this figure

Composition error in percentage points against time after a step. Without compensation the error reaches −4.5457 at about 40 s and stays outside ±0.1 until 412.0 s. With lead 90 s and lag 30 s the peak is −3.3308 and the error is inside ±0.1 from 109.5 s. Wild path: time constant 30 s, dead time 5 s; follower path: 90 s and 15 s; the 10 s dead-time difference remains. All parameters illustrative.

Interpreting the example correctly

The example demonstrates four practical lessons:

Equal final flow changes do not guarantee equal transient responses.
A ratio multiplier corrects steady-state proportion but not path dynamics.
Lead/lag compensation may reduce peak error and settling time when based on identified dynamics.
Dead time cannot be canceled safely by pretending the delayed stream has already responded.

The numerical settings are not general tuning rules. Real settings must come from approved tests or a validated process model. Dynamics may change with valve position, line velocity, equipment lineup, batch state, production rate, or filter settings.

Where compensation belongs

Dynamic compensation may be applied to the wild-flow signal before multiplication, to a demand signal, or within a broader feedforward model. Placement affects initialization, limits, tracking, and low-flow behavior.

Check:

whether the dynamic block tracks when bypassed;
its initial state after restart;
whether it acts on a valid linear-flow signal;
how it behaves when flow quality is bad;
whether limits are applied before or after compensation;
whether selector changes cause a discontinuity; and
whether its output can request an impossible controlled flow.

A filtered signal is not automatically a suitable lag element. Filter purpose, order, initialization, and location matter. Adding filters casually can worsen mismatch by delaying one path more than the other.

Verification after changes

After an approved compensation change, repeat comparable tests and compare:

peak downstream deviation;
time to enter and remain within a specified band;
controlled-flow setpoint demand;
valve movement;
saturation duration;
noise amplification;
behavior at low and high throughput; and
transfer behavior between manual, auto, and cascade.

Use the process acceptance criteria defined for the site. Do not convert the illustrative ±0.1 percentage-point analysis band into a universal criterion.

Modes, tracking, initialisation and windup

Do this first: Identify the currently applied signal at every handoff—ratio station, remote setpoint, controller output, override selector, and final element—and verify inactive blocks track it.

A correctly calculated ratio can be ineffective if the controlled-flow loop is not accepting its remote setpoint. Common states include:

manual output;
local automatic setpoint;
remote or cascade setpoint;
override-selected output;
initialization hold;
bad-input fallback; and
shutdown or permissive constraint.

Display labels vary by system. Confirm behavior from the logic and the actual selected path.

Manual, automatic, local, and remote modes

In manual, the operator or sequence directly sets output. In local automatic, the flow controller regulates to its internal setpoint. In remote or cascade mode, it accepts the ratio station’s setpoint. A cascade break may freeze, initialize, or replace that setpoint depending on configuration.

When troubleshooting, record both the requested and applied values. A ratio station may calculate 20 m³/h while the flow controller continues regulating to a local setpoint of 16 m³/h. Both blocks can appear healthy when viewed separately.

Bumpless transfer example

Consider ILLUSTRATIVE:

wild flow: 200 m³/h ILLUSTRATIVE;
tracked ratio: 0.04 ILLUSTRATIVE;
stale untracked ratio: 0.10 ILLUSTRATIVE.

The currently applied setpoint based on the tracked ratio is:

Tracked SP = 0.04 × 200 = 8 m³/h CALCULATED

If the ratio station switches in with the stale value:

Stale SP = 0.10 × 200 = 20 m³/h CALCULATED

The setpoint jump is:

Jump = 20 − 8 = 12 m³/h CALCULATED

Relative to the prior setpoint:

Jump percentage = 100 × 12 / 8 = 150% CALCULATED

A bumpless strategy should initialize or track inactive calculations so their outputs agree with the currently applied value before selection. Depending on architecture, this can require tracking:

ratio-station output;
remote setpoint;
ratio value;
bias;
trim integral;
lead/lag state;
controller integral term;
selector outputs; and
downstream applied output.

Tracking must use the applied or selected value, not merely an unselected request.

Integral windup

Integral windup occurs when controller integration continues while the applied output is constrained. Sources of constraint include valve limits, selector overrides, rate limits, interlocks, manual mode, and downstream saturation.

When the constraint clears, accumulated integral action can hold the output at a limit long after the process has crossed setpoint. The result may be severe overshoot and slow recovery.

An illustrative model uses ILLUSTRATIVE:

plant time constant: 30 s ILLUSTRATIVE;
PI proportional gain: 2 ILLUSTRATIVE;
integral gain: 0.08 per second ILLUSTRATIVE;
setpoint: 80% ILLUSTRATIVE;
initial PV: 20% ILLUSTRATIVE;
valve upper limit: 40% until 120 s, then 100% ILLUSTRATIVE;
back-calculation tracking time: 10 s ILLUSTRATIVE.

A back-calculation form is:

I_dot = Ki e + (u_sat − u_unsat) / Tt

In the illustrative calculation, peak PV without anti-windup is 99.99%, an overshoot of about 20 percentage points CALCULATED. With back-calculation, peak PV is 80.00% CALCULATED.

After the valve limit is released at 120 s, the calculated time, measured from that release (not from t = 0), to enter and remain within ±1 percentage point of setpoint is 359 s without anti-windup and 88 s with it CALCULATED. The model is a first-order plant (τ = 30 s) with a PI controller, a 0.2 s Euler step, and initial PV 20 %.

These results belong to this model. They are not guaranteed outcomes or recommended settings. The Notre Dame instructional notebook on integral windup and bumpless transfer provides educational background SOURCE.

Figure V10 shows how the unsaturated output continues accumulating while the applied output is limited, and how back-calculation uses the applied saturated value.

Figure V10. Anti-windup limits integral accumulation and brings the response back within ±1 point much sooner after the valve limit is released (illustrative). Illustrative model — calculated

Text alternative for this figure

Plant time constant 30 s, PI Kp 2 and Ki 0.08 per second, set point 80 %, valve limit 40 % until 120 s then 100 %. Without anti-windup the PV peaks at 99.99 % and takes 359 s after the release to stay within ±1 point; with back-calculation anti-windup it peaks at 80.00 % and takes 88 s.

What to inspect in a live strategy

Trend these variables separately if available:

proportional contribution;
integral contribution;
unsaturated output;
output after limits;
output after selectors;
output delivered to the final element;
controller mode;
track status;
saturation flags;
setpoint source;
local and remote setpoints; and
initialization status.

A controller may include built-in anti-windup but still wind up relative to an external selector if the block is not told which output was actually selected. Override schemes need external-reset feedback, output tracking, or the platform’s equivalent mechanism.

Transfer test

An approved transfer test should verify:

The inactive source tracks the applied value.
Transfer causes no unexpected setpoint step.
Output remains within limits.
The valve moves in the expected direction.
Returning to the original source is also bumpless.
Trim and bias states remain valid.
Bad-input and restart behavior are defined.
Saturation flags clear when the constraint clears.
Operator displays identify the selected source.

Ratio trim, bias and analyzer delays

Do this first: Put the inner flow loops in a verified condition before diagnosing or tuning analyzer trim.

Ratio trim is a slow correction around a faster base ratio. It can compensate for gradual changes in composition, material properties, or process efficiency when the measured result provides reliable feedback. It should not be used to conceal incorrect flow scaling, a saturated valve, or a loop left in manual.

Bias versus trim

A bias adds a fixed term to a flow setpoint:

SP = R × F_wild + B

A ratio trim changes the proportional relationship, for example:

R_applied = R_base × Trim_factor

or:

R_applied = R_base + Trim_offset

These forms are not equivalent. A fixed flow bias has a larger relative effect at low throughput. A multiplicative ratio trim maintains the same fractional correction across throughput, subject to its limits.

Document:

base ratio;
trim form;
trim units;
minimum and maximum trim;
output limits;
analyzer setpoint;
analyzer direction;
quality handling;
initialization; and
behavior when the analyzer is unavailable.

Symptom distinctions

Symptom	Important distinction	Test direction
Trim at a limit with residual error	Insufficient authority, wrong base ratio, capacity constraint, or analyzer bias	Verify base calculation, analyzer calibration, inner-loop tracking, and physical limits
Trim cycles while inner flows track	Analyzer delay, excessive loop gain, filtering, or interaction	Trend analyzer sample timing and hold trim in an approved fixed state
Inner controlled flow does not track	Measurement, valve, saturation, or mode problem	Repair the inner flow loop before adjusting trim
Trim changes but ratio setpoint does not	Selection, limit, scaling, or disabled path	Trace the applied ratio and remote-setpoint path
Correct analyzer value but wrong product result	Sampling location, representativeness, laboratory comparison, or unmeasured disturbance	Verify sample system and process basis
Trim initializes at the wrong value	Tracking or restart-state problem	Compare stored state with the currently applied ratio

Bounded authority

Trim limits prevent a slow supervisory loop from demanding an unlimited ratio change. Hitting a limit is diagnostic information. It can indicate:

the base ratio is wrong;
process conditions moved beyond the intended range;
the controlled-flow valve lacks capacity;
a flow meter is biased;
the analyzer is biased;
the analyzer setpoint is inappropriate;
an unmeasured disturbance is dominant; or
trim authority is insufficient for the approved operating case.

Do not increase authority simply because trim is at a limit. First identify why it reached the limit and whether a larger correction is safe and physically meaningful.

Analyzer delay

Analyzer feedback may contain:

transport delay from sample point to analyzer;
sample conditioning delay;
cyclic measurement time;
validation delay;
digital filtering;
laboratory comparison delay; and
dead volume between blending and sample location.

These delays can make a trim loop oscillate even while both flow loops track perfectly. The analyzer correction arrives after the condition that caused it has changed. Additional gain or integral action can make the cycle worse.

There is no universal trim-to-flow-loop tuning ratio. Determine settings from measured dynamics, analyzer quality, operating objectives, and approved site practice. Tune and validate the inner flow loops first. Then test the analyzer path and apply bounded trim slowly enough for its actual delay and process response.

Analyzer verification checklist

Before changing trim tuning:

Confirm analyzer calibration and maintenance status.
Compare with the approved independent reference method.
Verify sample-point identity.
Measure or estimate sample transport time through an approved test.
Check sample flow, pressure, temperature, and conditioning.
Record analyzer cycle and validation time.
Identify filtering in the analyzer and control system.
Confirm bad-quality and maintenance-mode behavior.
Check whether the trim freezes, tracks, or resets when quality is bad.
Verify trim limits and direction.
Confirm inner flow loops track during the entire observation.
Repeat observations at relevant operating points.

Combustion cross-limiting

Do this first: Treat the combustion control narrative, cause-and-effect documentation, burner-management requirements, and approved operating procedure as controlling documents.

Simple fuel-to-air ratio control can produce a transient shortage of air when firing demand rises or excess fuel when demand falls. Cross-limiting coordinates air and fuel setpoints so air leads during an increase and fuel leads during a decrease.

Practitioner and vendor examples from ControlGuru and MG Co. describe this principle SOURCE. These examples are educational and illustrative, not prescriptive designs for a particular boiler, furnace, or burner.

A simplified arrangement can use:

a high selector for air demand, so air demand cannot fall below the amount required by measured fuel; and
a low selector for fuel demand, so fuel demand cannot rise above the amount permitted by measured air.

Increases are constrained until sufficient air is measured. Decreases reduce fuel first while air follows. Actual systems may include characterization, oxygen trim, pressure control, minimum-air logic, purge states, multiple fuels, fan constraints, limits, and protective systems.

Illustrative normalized simulation

Consider a normalized simulation ILLUSTRATIVE:

demand change: 50% to 80% ILLUSTRATIVE;
air margin: 10% ILLUSTRATIVE;
air-path time constant: 8 s ILLUSTRATIVE;
fuel-path time constant: 5 s ILLUSTRATIVE;
initial fuel: 50 ILLUSTRATIVE;
initial air: 55 ILLUSTRATIVE.

The cross-limited setpoints are:

AirSP = max(1.10 × demand, 1.10 × measured fuel)

FuelSP = min(demand, measured air / 1.10)

The normalized excess-air indicator is:

EA (%) = 100 × (air / fuel − 1)

In this simulation, the minimum EA indicator is 10.0% with cross-limiting and 1.92% with basic ratio control CALCULATED. These are normalized, dimensionless simulation values. They are not measured oxygen, not stoichiometric guidance, not a safety limit, not a design margin, not evidence of acceptable combustion or compliance, and not recommended settings, plant data or universal excess-air targets.

Time (s) ILLUSTRATIVE	Cross-limited air CALCULATED	Cross-limited fuel CALCULATED	Cross-limited EA (%) CALCULATED	Basic air CALCULATED	Basic fuel CALCULATED	Basic EA (%) CALCULATED
0	55.41	50.00	10.83	55.41	50.60	9.51
8	76.09	60.85	25.03	76.09	74.16	2.60
16	83.65	71.38	17.19	83.65	78.84	6.10
24	86.41	76.53	12.92	86.41	79.77	8.32
32	87.42	78.67	11.13	87.42	79.95	9.34
40	87.79	79.50	10.43	87.79	79.99	9.75

Figure V09 shows the normalized trajectories and how the selectors hold fuel demand behind measured air during the load increase.

Figure V09. Illustrative cross-limiting keeps fuel demand from outrunning the available air during a load increase; it is not a burner-management design. Illustrative — not a BMS design

Text alternative for this figure

Air demand = load demand × 1.10, then a high selector with measured fuel × 1.10, then the air flow PID and damper. Fuel demand passes a low selector with measured air ÷ 1.10, then the fuel flow PID and valve. For a load rise from 50 to 80 %, air leads and fuel follows. Minimum normalised excess-air indicator: 10.0 % with cross-limiting, 1.92 % with basic ratio control. Illustrative control logic, not a burner-management system design.

How to troubleshoot cross-limiting

Start with the selected signals, not only the controller faceplates:

Record demand, measured air, measured fuel, calculated constraints, and selector outputs.
Identify which input each selector has selected.
Verify air and fuel measurements use their intended characterized bases.
Check tracking of unselected demand paths.
Confirm limits and rate constraints are active as documented.
Compare setpoints with measured responses.
Check actuator saturation and position feedback.
Verify transitions among startup, purge, light-off, normal firing, low-fire, and shutdown only through approved procedures.
Confirm alarms, interlocks, and burner-management functions remain independent and effective.
Restore and document the approved configuration.

Drawing interpretation may also require tracing combustion-air and fuel-gas services. The overview of utility flow diagrams and P&IDs provides general context, while the project drawings and cause-and-effect documents govern the installation.

Commissioning test procedure and checklist

Do this first: Agree on the test plan, responsible roles, operating limits, rollback conditions, and required records before applying any change.

Ratio-loop commissioning should prove the full signal path rather than only the controller response. Use approved site procedures and vendor documentation. For drawing-navigation fundamentals, see the beginner guide to reading a P&ID and the comparison of PFD and P&ID information.

Commissioning record

Record the following:

Raw wild-flow transmitter input.
Scaled wild-flow PV.
Raw controlled-flow transmitter input.
Scaled controlled-flow PV.
Square-root status and exact processing location.
Wild flow in engineering units.
Selected base ratio and ratio definition.
Bias value and units.
Trim value, form, authority, and limits.
Calculated controlled-flow setpoint.
Setpoint after limits and selectors.
Setpoint received by the controlled-flow controller.
Controlled-flow PV.
Controller command.
Output after saturation limits.
Output after selectors or overrides.
Valve or drive command.
Actual stem position or speed feedback.
Active modes for every block.
Tracking and initialization flags.
Saturation and constraint flags.
Transmitter, controller, and output ranges.
Density and composition assumptions.
Gas reference temperature and absolute pressure where applicable.
Final-element fail action.
Output and rate limits.
Step-test starting conditions.
Step magnitude, direction, and authorization.
Measured dead time and response time.
Transfer response among manual, auto, local, and remote modes.
Low-flow cutoff, hysteresis, and recovery behavior.
Zero-flow and shutdown behavior.
Bad-quality behavior for every critical input.
Restart and initialization behavior.
Analyzer delay, quality handling, and trim response.
Restoration status and final approved configuration.
For combustion, proof-test through the approved procedure; ratio commissioning does not validate the burner-management system.

Functional test sequence

Verify documentation, calibration status, ranges, units, and signal quality.
Prove the wild-flow signal from input to ratio station.
Prove the controlled-flow signal from input to controller.
Verify exactly one required square root in each DP flow path.
Test ratio arithmetic at several approved values.
Verify bias and trim equations independently.
Test limits, selectors, and quality handling.
Verify remote-setpoint transfer and tracking.
Test final-element direction and feedback.
Conduct approved small response tests.
Confirm saturation and anti-windup behavior.
Test low-flow, zero-flow, shutdown, and restart behavior.
Verify analyzer trim only after the inner loops pass.
Restore all normal modes and document the final state.

Troubleshooting resources

Useful resources include calibrated test instrumentation appropriate to the installed signals, access to synchronized trends, approved configuration records, current drawings, control narratives, valve data, transmitter records, and authorized calculation templates.

The downloadable checklist should be completed with actual tag names, ranges, units, mode terminology, and site approval fields. A generic checklist cannot replace a plant procedure.

Final fault-isolation checklist

Before closing the investigation, confirm:

The ratio definition is documented.
The multiplier equation is correct.
Signal units and bases are compatible.
Span conversion is correct.
Square-root processing occurs exactly once where required.
Low-flow behavior is defined.
The generated setpoint matches an independent calculation.
The controlled loop is accepting the remote setpoint.
Inactive blocks track the applied value.
The controller is not unintentionally saturated.
Selected output reaches the final element.
Stem feedback follows command.
The valve has adequate installed controllable range.
Bypass and seat leakage have been considered.
Dynamic mismatch has been tested separately.
Lead/lag settings are supported by measured dynamics.
Anti-windup uses the applied or selected output.
Trim is bounded and not masking an inner-loop fault.
Analyzer delay and calibration are known.
All protective functions remain intact.
The approved configuration has been restored and recorded.

Frequently Asked Questions

Do this first: Match the question to the correct layer—architecture, measurement, final element, dynamics, or supervisory trim—before selecting a remedy.

What is the difference between ratio control and feedforward control?

Ratio control is a specific feedforward arrangement in which a measured wild variable generates another loop’s setpoint through a defined proportional relationship. General feedforward control may use several disturbances, nonlinear models, dynamic compensation, and additional process variables. Ratio control is therefore a subset or structured form of feedforward control rather than a completely separate principle.

Why not control the calculated ratio directly with a PID?

A divided ratio F_controlled/F_wild becomes extremely sensitive as wild flow approaches zero. Noise, offsets, and leakage can then create very large apparent ratio errors and aggressive controller action. Multiplying wild flow by the required ratio to generate a controlled-flow setpoint avoids making division the primary feedback variable.

Should I use ratio on flow or on measured values such as analyzer composition?

Use flow ratio for fast material-rate coordination when flow is the appropriate physical basis. A slower analyzer can trim the base ratio when composition feedback is reliable, but it should not replace functioning inner flow loops. Analyzer delay, filtering, calibration, and sample transport must be considered explicitly.

Why is my ratio inaccurate at low flow?

Possible causes include operating below useful meter range, differential-pressure square-root sensitivity, zero error, leakage, cutoff behavior, quantization, and division by a small wild-flow value. Inspect raw measurements and validity logic before adjusting the ratio. A stable display created by filtering does not prove the underlying low-flow measurement is accurate.

What is ratio bias, and when is it used?

Bias is a fixed additive term, as in SP=R× F_wild+B. It may represent a documented fixed contribution, but it causes the instantaneous controlled-to-wild ratio to vary with throughput. Its units, low-flow behavior, limits, and process justification must be explicit.

Why does the ratio jump when I switch to auto?

The inactive ratio station, bias, trim, remote setpoint, or controller integral state may not have tracked the applied value. When selected, the stale value becomes active and creates a step. Verify initialization and tracking at every selector and mode handoff, including the applied final output.

Does ratio control need density compensation?

It needs density compensation when the required relationship is on a mass basis but one or both measurements are volumetric and density changes materially. If the specification is genuinely volumetric under defined conditions, density compensation may not be required. For gases, document reference temperature and absolute pressure for standard-volume signals.

What causes oscillation in a ratio loop?

Potential causes include aggressive inner-loop tuning, valve stiction, saturation and windup, measurement noise, duplicate filtering, interacting flow loops, analyzer delay, excessive trim gain, and selector switching. Trend setpoints, PVs, outputs, stem feedback, modes, and selected signals together to distinguish them.

How is cross-limiting different from simple ratio control?

Simple ratio control calculates one setpoint as a proportional function of another flow or demand. Cross-limiting adds selectors so air leads fuel on an increase and fuel leads air on a decrease. It coordinates combustion demands but does not replace burner-management protection, interlocks, or approved combustion procedures.

Limits of This Guide

Do this first: Compare this general guidance with the site’s approved procedures, control narrative, vendor documentation, and applicable requirements.

This guide is educational and is not a substitute for site procedures, vendor instructions, hazard review, engineering approval, or professional judgment.
The worked examples are illustrative calculations, not plant data.
The screening table cannot diagnose dynamic, mechanical, density, or analyzer faults from one snapshot.
Numerical controller, lead/lag, cutoff, trim, and stiction examples are not universal settings or acceptance limits.
The guide does not provide burner-management test procedures.
Cross-limiting examples do not establish combustion targets, stoichiometric ratios, flue-gas oxygen settings, or excess-air requirements.
Valve performance depends on the installed process system, not only inherent catalog characteristics.
Measurement performance depends on technology, installation, operating conditions, range, compensation, and maintenance.
Dynamic compensation must be based on approved tests or a validated model.
Safety instrumented functions, interlocks, trips, alarms, and regulatory duties require their own approved lifecycle and proof-test procedures.
Site-specific tag conventions, mode names, block behavior, and quality handling may differ from the terminology used here.
Changes to live control systems require authorization, version control, testing, rollback planning, and restoration records.

Sources and Further Reading

Do this first: Use these pages for background, then verify every implementation detail against the actual project documents and configured system.

The external source pages below are educational, practitioner, standards-scope, or vendor material of varying authority. They support general principles but do not replace project specifications, current standards, approved procedures, or equipment documentation.

LibreTexts: Ratio control—what it is, when useful, and common usage/11%3A_Control_Architectures/11.04%3A_Ratio_control-_What_is_it_When_useful_When_not_Common_usage) — educational overview of ratio-control architecture.
Control.com: Ratio control — educational explanation of ratio strategies.
ControlGuru: The ratio control architecture — practitioner explanation of multiplier and setpoint architecture.
ControlGuru: Ratio with cross-limiting override control of a combustion process — practitioner combustion example.
MG Co.: Cross-limiting combustion control example — vendor application example.
Yokogawa: Low-flow cutoff — product-specific application information.
Control.com: Flow measurement from differential pressure — educational square-root explanation.
ISA-75.11 committee page — standards scope information related to control-valve inherent flow characteristic and rangeability.
Notre Dame instructional notebook: Integral windup and bumpless transfer — educational control example.
NFPA 85 — code scope reference for covered combustion systems.

For project drawing literacy and supporting concepts:

Disclosure and Review Status

This draft contains educational engineering guidance and illustrative calculations. It does not replace approved plant procedures, vendor documentation, applicable codes, site-specific risk assessment, or qualified professional approval. Any commercial relationship associated with an affiliate marker must be disclosed by the publisher clearly and close to the relevant content.