Dry gas seal support-system monitoring

DryGasSealMonitor evaluates caller-normalized measurements for one tandem seal. Use separate instances for each end of the compressor. It produces fault evidence and simulated trip recommendations; it does not send commands or certify a protection function.

Measurements, limits and interpretation

Signal	Reference/basis	Observable concern
Supply minus process pressure	Both absolute bara; difference in bar	Insufficient pressure barrier, possible process ingress
Primary vent flow	Ideal NL/min at 273.15 K and 1.01325 bara	Excessive leakage or another vent contribution
Primary vent pressure	Absolute bara	Backpressure or restricted disposal path
Intermediate buffer gas flow	Same normal-volume basis	Insufficient buffer supply
Separation differential pressure	Separation pressure minus bearing cavity pressure, bar	Inadequate oil/process migration barrier
Gas-quality assessment	Explicit externally evaluated pass flag	Wet/contaminated gas or unresolved quality
Instrument quality and oldest sample age	Boolean quality, seconds	Missing, stale or physically invalid measurements

Do not substitute supply-to-primary-vent differential pressure for supply-to-process pressure, or confuse intermediate buffer gas with bearing-side separation gas. A negative differential pressure is a physical barrier fault; negative absolute pressure or normal flow is invalid measurement evidence. Root causes are hypotheses: a high vent-flow signal alone does not uniquely diagnose a failed primary seal.

All Limits values are supplied by the caller/OEM. No universal leakage, pressure or trip-delay thresholds are assumed. The primary vent maximum pressure must be positive; all other limits must be finite and nonnegative. A zero confirmation delay requests an immediate recommendation for an active physical fault.

Timers and replay

evaluate(sample, intervalSeconds) represents one non-overlapping scan interval. Use a finite positive interval in seconds. Physical faults must persist continuously for the configured confirmation delay. Healthy scans reset the corresponding timer. Invalid scans break continuous physical-fault evidence and immediately latch INVALID_DATA. The caller therefore receives unresolved-data evidence rather than an invented healthy state.

Confirmed recommendations remain latched until reset(). Reset clears all timers and recommendations; an ongoing fault reappears on the next scan. Reset is an explicit simulator operation, not a plant reset permissive. Results are immutable detached snapshots, and serialization preserves the monitor timers/latches for replay. Each collection getter returns an unmodifiable defensive copy in fault declaration order. Retained collections and serialized results remain unchanged by later scans or monitor resets.

This initial monitor has one common confirmation delay and a single running/pressurized measurement basis. Mode-specific source/booster switching, alarm/trip threshold separation, hysteresis, startup/standstill/coastdown logic, sensor voting and complete controller/SIS integration remain acceptance items in issue #4145.

Example (synthetic limits)

DryGasSealMonitor.Limits limits =
    new DryGasSealMonitor.Limits(2, 100, 3, 10, 0.1, 2, 3);
DryGasSealMonitor monitor = new DryGasSealMonitor(limits);
DryGasSealMonitor.Sample sample =
    new DryGasSealMonitor.Sample(82, 80, 101, 2, 10, 0.1, 0, true, true);
DryGasSealMonitor.Result first = monitor.evaluate(sample, 2);
DryGasSealMonitor.Result confirmed = monitor.evaluate(sample, 1);

The example’s high vent flow lasts three seconds and confirms a recommendation on the second scan. The numbers illustrate behavior; use project-specific limits. This API sequence is exercised by DryGasSealMonitorTest.delayRecoveryLatchAndReset. The tests also inject simultaneous faults, bad/stale measurements, interrupted evidence, recovery/reset and serialization/replay.

See the program roadmap #4149, EagleBurgmann seal management systems and the bearing-side separation barrier description.