Stage 1 · TEST

Define the loop and validate the test

Import a step-test dataset, map its signals, inspect data quality, and confirm a useful manipulated-variable step before any tuning is calculated.

Loop identity

Required

Loop tag

Loop category

Equipment / system

Location

Process variable

PV units

Manipulated variable

Output units

Controller action

Controller type

Controller form

Execution interval

s

Normal operating range

Controller manufacturer

Controller model

Test date

Analyst

Reviewer

Test constraints

Import baseline data

Drop CSV, TSV, or JSON here

Core import works offline. Original rows are retained.

Choose file

No baseline dataset loaded.

Signal mapping

Map the imported columns. Time, process variable, and manipulated variable are required.

Awaiting data

Time

Setpoint

Process variable

Controller output

Controller mode

Valve feedback

DATA LAB

Reversible preparation, exclusions, annotations, and event-window control.

Analysis filter

Filter window

samples

Isolated spike rejection

σ (0 = off)

Analysis start

s

Analysis end

s

Resample interval

s (0 = off)

Step sensitivity

Exclusion regions

Removed from calculations

Start

End

Reason

Annotations

Evidence retained

Time

Type

Label

Selected event editor

Adjust the step, baseline, and response windows without modifying source rows.

No event

Step time

s

Baseline start

s

Baseline end

s

Response end

s

Original imported values remain retained and can be restored.No preparation applied

Test quality and candidate steps

The selected event controls model estimation.

Not evaluated

Status

No data

Candidate steps will appear after mapping.

Wheel to zoom · drag to pan · drag the amber step marker to adjust the event.No data

Stage 2 · ESTIMATE

Estimate process behavior

Compare threshold, reaction-curve, and numerical fits. The selected practical model remains visible with its uncertainty and reasons it may be unreliable.

Analyze a mapped test before estimating a model.

MODEL LAB selection

Automatic ranking uses fit, parameter count, residual behavior, and process-shape evidence. A manual selection is recorded.

Auditable

Selected model

No model

No model estimate.

Alternative estimates

Transparent comparison

No alternative estimates.

Residual and uncertainty diagnostics

No diagnostics

No residual diagnostics.

Model output is a characterization, not a guarantee of field behavior.—

Stage 3 · CHARACTERIZE

Interpret the loop before tuning

LOOP PERSONALITY uses deterministic evidence, counter-evidence, and measured indicators. Severe conditions take precedence over routine tuning.

Signature feature

LOOP PERSONALITY

Awaiting analysis

No classifications available.

v1.4 · DEADBAND LAB

Deadband and hysteresis

Not evaluated

Bidirectional evidence has not been evaluated.

Instability and constraint warnings

No review

Warnings will appear after analysis.

DEADBAND LAB evidence

Staircase events, direction thresholds, reversal behavior, quantization, actuator feedback, and operating-point variation are evaluated separately.

No nonlinear evidence available.

Likely first action

No action recommendation.

Recommended additional test

No procedure generated.

Diagnostic domains

Root-cause orientation

No diagnostic domains.

Classification conflicts

Precedence review

No conflicts evaluated.

Why not another personality?

Every supported rule shows its threshold, evidence status, and reason for assignment or rejection.

No rule evaluations.

Event-to-event behavior

Operating-point evidence

No event profiles.

Classification override

Recorded hypothesis

Personality

Action

Reason

Flat response alone is not labeled deadband; noise, resolution, delay, and excitation are considered.

Points are ordered by event time; loop area is a diagnostic indicator, not proof of a single fault.

Valve feedback is optional. Missing feedback prevents a direct stiction or breakaway assessment.

Stage 4 · TUNE

Generate guarded, conservative guidance

Recommendations are model- and controller-form aware. Calculated values, guardrails, formulas, assumptions, warnings, and rollback settings remain visible together.

A suitable or explicitly overridden test and a reliable model are required.

v1.5 · CONTROLLER TRANSLATOR

Current controller settings

Rollback source

Displayed proportional value

Displayed integral value

Displayed derivative value

Derivative filter N

Conservatism and limits

Robustness first

Closed-loop target

User target λ

s

PV low safety limit

PV high safety limit

Max gain increase

%

Max gain decrease

%

Max integral-time change

%

Output minimum

%

Output maximum

%

Override unsuitable test

Controller form and unit translation

The application converts displayed settings to a canonical ideal PID form, performs calculations there, and converts the recommendation back with an audit trail.

Not checked

Controller profile

Proportional convention

Integral convention

Derivative convention

Derivative signal

Implementation

Configured PV span

Configured output span

No conversion audit available.

v1.6 · TUNING WORKBENCH

Control objective and implementation guidance

Conservative default

Primary objective

Tuning family

Extra robustness margin

%

Max derivative-time change

%

Anti-windup status

Setpoint filter

Split-range output

Selected scenario

No tuning scenarios available.

No implementation guidance available.

Tuning comparison

No recommendation

Generate a recommendation after analysis.

User-modified preview

Simulation only

Canonical Kp

Canonical Ti

s

Canonical Td

s

Simulation only. User-modified values are not recommendations and do not bypass warnings, approval, rollback, or field authorization.

Measured retest evidence always takes precedence over this preview.—

Tuning card

No tuning card available.

Step-test procedure

Generate analysis to create a tailored procedure.

Stage 5 · RETEST LAB

Compare repeated field evidence

Retain multiple named retests, verify comparable operating conditions, apply multi-metric acceptance criteria, and document response tradeoffs and personality transitions.

v1.7 · RETEST LAB

Retest register

No retests registered.

Active retest evidence

Retest name

Operating condition

Retest date

Disposition

Retest notes

Drop retest CSV, TSV, or JSON here

Baseline mapping is reused when matching columns are found. Source timestamps remain retained.

Choose retest file

No retest loaded.

Comparison result

Not compared

Import and compare a retest.

Acceptance criteria

Maximum settling time

s

Maximum overshoot

%

Maximum normalized IAE

Maximum output travel

Performance metrics

Multiple criteria

No comparison metrics.

Cross-retest scorecard

All retained tests

No completed retests.

Original timestamps remain in each retained dataset; plot traces align their selected step at zero.

v1.8 · REPORT BUILDER

Assemble the issued evidence package

Choose report depth, document-control fields, included evidence sections, conclusions, and approval status before previewing or exporting.

Document control

Draft

Report type

Report title

Document ID

Revision

Status

Prepared for

Purpose and scope

Conclusions and disposition

Included sections

Approvals

Role

Name

Report readiness

Not reviewed

Preview outline

v1.9 · PROJECT OPERATIONS

Manage the local project library

IndexedDB stores multiple projects, archive state, recovery snapshots, migrations, and project metadata without sending data to a server.

Current project

Local

Revision difference

Current vs latest snapshot

No recovery snapshot.

Local project library

Search

Show

Library is loading.

Migration and storage log

v2.0 project format

ADVANCED

Revision history

Overrides, warning acknowledgments, model selections, tuning changes, and project actions are recorded locally.

No revisions recorded.