delphi-pump-analytics / analysis /knowledge /operational_context.yaml
pandeydigant31's picture
Initial deploy: DELPHI Pump Analytics (renamed from MURPHY)
199bfa3 verified
Raw
History Blame Contribute Delete
7.07 kB
# CSH2 Cryogenic Pump — Operational Context Knowledge
# Practical knowledge that operators have but specifications don't document.
# Derived from maintenance_log.xlsx, Technical PDF, and data analysis.
sensor_trustworthiness:
TT110:
notes: >
TT110 reads pipe wall temperature, not fluid temperature. After cooldown,
it needs 10-15 minutes to equilibrate with the flowing cryogen. During
transients (pump start/stop, flow changes), TT110 lags the actual fluid
temperature by 5-30 seconds depending on flow rate.
workaround: >
For precise inlet temperature, use virtual T_inlet back-calculated from
discharge state via ThermoProps (physics/thermo.py). The virtual sensor
uses PT110 + TT130 + isentropic relations to infer inlet conditions.
known_faults:
- "Jul 16, 2025: 0-32K readings are sensor fault, not real cryo temps"
- "Filter out readings during known warm periods where TT110 < 50K"
FT140:
notes: >
FT140 is a Coriolis mass flow meter calibrated for LN2. When running LH2,
the density difference (LN2: 808 kg/m³ vs LH2: 70.8 kg/m³) causes the
raw reading to be 100-200x too high. The FT140Calibrator applies a density
correction factor of approximately 11.4x.
workaround: >
Always apply FT140Calibrator (physics/ft140_calibration.py) density correction
before interpreting flow readings. Near-zero flow at low pump speeds is
expected and not a fault — the meter has a low-flow cutoff.
known_faults:
- "Raw FT140 reads ~100-200x too high for LH2 — apply correction"
- "Near-zero flow at idle is expected, not a sensor fault"
PT100:
notes: >
PT100 is stuck at -14.86 bar in the current dataset. This is not a
physical pressure — it's a sensor fault.
workaround: >
Use default_when_invalid=2.0 barg for any calculations requiring PT100
(e.g., Psat estimation). This is the saturation pressure approximation.
known_faults:
- "Stuck at -14.86 bar — use 2.0 barg default"
CalcMass:
notes: >
CalcMass is stuck at 0.18 kg in the current dataset. The PLC calculation
is not updating.
workaround: >
Use FT140 integration (corrected) for actual dispensed mass calculation.
Integrate corrected FT140 over the fill duration.
known_faults:
- "Stuck at 0.18 kg — use FT140 integration instead"
M130_Speed_vs_MC130:
notes: >
Tag 68 (M130_Speed) replaces tag 13 (MC130_VFD_Speed) after May 28, 2025.
Both map to motor speed percentage but are different register addresses.
TAG_ALIASES in config.py handles this mapping.
workaround: >
Check TAG_ALIASES. Use tag 68 for post-May data, tag 13 for pre-May.
Both are valid speed readings.
post_maintenance_expectations:
hp_seal_replacement:
description: "After HP seal replacement or energizer work"
expected_behavior:
- "Initial friction: 7-9 kgf (from Jul 18 maintenance entry)"
- "Break-in period: ~1000 cycles to reach 4-5 kgf steady state"
- "Motor current will be higher initially — this is normal"
- "Volumetric efficiency may be slightly higher initially (tighter seal)"
timeline: "1000 cycles (~2-3 days at normal testing rate)"
ref: "2025-07-18 maintenance event"
icv_replacement:
description: "After ICV replacement or reseating"
expected_behavior:
- "Immediate improvement in pressure build if ICV was the issue"
- "New valve may need seating-in (initial performance varies)"
- "Jul 22: orientation fix restored 700 bar immediately"
- "Jul 14: polished ICV helped but didn't fully fix — surface not flat"
timeline: "Immediate improvement expected"
ref: "2025-07-01, 2025-07-14, 2025-07-22 events"
dcv_polishing:
description: "After DCV sealing surface polishing"
expected_behavior:
- "Immediate improvement in leak test results"
- "Run PT130 decay test before and after to quantify improvement"
- "Jul 1: polishing improved seal but some residual leakage"
timeline: "Immediate improvement"
ref: "2025-07-01 maintenance event"
regime_caveats:
ln2_vs_lh2:
description: "LN2 and LH2 data are not directly comparable"
details: >
LN2 (77K, 808 kg/m³) and LH2 (20K, 70.8 kg/m³) have very different
fluid properties that shift all performance curves. Compression ratios,
volumetric efficiencies, heat transfer rates, and valve dynamics all
change significantly between fluids. The SPEED_SCALE_FACTOR in config.py
was calibrated for LN2 and needs adjustment for LH2.
impact:
- "Flow rate scales with density ratio"
- "Pressure build rate changes with bulk modulus"
- "Thermal effects are very different (77K vs 20K)"
pressure_drift:
description: "Discharge pressure drifts down during extended fills"
details: >
Documented in Technical PDF p.22. The pump operates at approximately
fixed pressure ratio. As the small test cryotank depletes, inlet
pressure (PT110) drops, and PT130 follows. This is NOT a fault —
it's an expected consequence of the small tank size.
impact:
- "PT130 and flow rate decline over extended fills"
- "Production tanks will be sized to maintain pressure"
subcooling_sensitivity:
description: "Pump performance is very sensitive to inlet subcooling"
details: >
From Technical PDF p.17: the pump is 169x more efficient with
subcooled liquid than with gas. Even small amounts of flash gas
at the inlet significantly reduce volumetric efficiency. Ensure
adequate subcooling margin (>5K) before starting fills.
impact:
- "Inlet quality > 0 = reduced efficiency"
- "Subcooling margin < 2K = risk zone"
system_personality:
recurring_issues:
- priority: 1
description: "ICV sealing is the #1 recurring issue"
details: "3 occurrences in July 2025. Check ICV first for any pressure build problem."
refs: ["2025-07-01", "2025-07-14", "2025-07-22"]
- priority: 2
description: "DCV sealing is the #2 issue"
details: "1 occurrence. DCV polishing was effective but residual leakage remained."
refs: ["2025-07-01"]
- priority: 3
description: "Safety loop tuning needs attention at high pressures"
details: "AOV141 hunting observed. PID tuning required."
refs: ["2025-07-22"]
strengths:
- "HP seal life is excellent (8.2M cycle lower bound = 3+ months at 20 fills/day)"
- "Phase 1B achieved 100% uptime across 6 fills"
- "Specific energy of 0.26 kWh/kg matches target"
- "Back-to-back fill capability demonstrated (4 consecutive fills)"
known_limitations:
- "Pump gas ratio 14.4% vs 2.7% target — efficiency gap exists"
- "Peak pressure 370 bar vs 700 bar H70 target — further testing needed"
- "Small test tank causes pressure drift during extended fills"
- "FT140 requires density correction for LH2 — raw readings are wrong"