Publish Time: 2025-09-16 Origin: Site
Rainy season approaches. We want extra lead time, fewer surprises, smoother shifts. This guide delivers a fast ten‑point pass in five minutes. It fits busy crews. It builds confidence before storms arrive. We keep it practical. We keep a friendly voice. We lean on KJ117(A) for trends and alerts. We pair screens and field steps. We finish strong on logging and backup.
You can copy the tables into a binder or a phone. You can turn the mini charts into small stickers near control panels. You can brief a new operator in one break. That saves hours later. Oops — we promised no "that." Let's keep going, no slipups.
Rainfall lag. Skies clear. Inflow keeps rising for hours. Sometimes for days.
Goaf water loves slow surprises. A gentle climb creeps in overnight. Then pumps play catch‑up.
Duty load grows in wet months. A short pre‑check frees attention for real problems.
Small fixes early. Big headaches avoided later.
Open KJ117(A). Scan three curves: water level, pressure, flow.
Spot‑check three hot spots underground: bends, pump‑room return, goaf boundary.
Send one mobile alert test. See every on‑duty phone confirm.
Run a 3‑minute pump test. Note flow and head. Listen for noise.
Log results. Push to cloud storage. Keep a local copy on a safe drive.
Mini flow map:
Trends → Blind-spot check → Alert test → Pump run → Log & archive
Use this sheet during the walk‑through. Keep it near the dispatch desk. Print large. Add check boxes.
| # | What to do | Pass rule | Frequent pitfall | Owner | Time |
|---|---|---|---|---|---|
| 1 | Coverage and blind spots at bends / returns / goaf edge | Critical points ≥ plan, neighboring curves cross‑validate | Only dashboard view, no local slice | Monitoring | 30–60 s |
| 2 | Power and UPS endurance | Auto recovery in ≤3 min after power cut, UPS ≥30 min | No power‑off drill ever | Mech/Elec | 30–60 s |
| 3 | Comms health on wired + radio links | Offline ratio <1%, key nodes online ≥30 days | Wet joints ignored | Monitoring/IT | 30–60 s |
| 4 | Sensor calibration and time sync | Drift within limits, time skew <2 s | Calibrate hardware, forget clocks | Maintenance | 30–60 s |
| 5 | Tiered thresholds and call tree | Lower false alarms, earlier warning ≥15 min | One single threshold for all | Monitoring | 30–60 s |
| 6 | Mobile alert delivery test | Every on‑duty person confirms in 3 min | Old contact list | Dispatch/EHS | 30–60 s |
| 7 | Pump test + redundancy check | Flow ≥ design × 1.2, standby pump ready | Main pump only, no swap test | Mech/Elec | 60–90 s |
| 8 | "Go‑bag" essentials in pump room | Five items present, reachable in 5 min | No labels, no inventory cycle | Mech/Store | 30 s |
| 9 | Goaf risk reinforcement | Higher sampling near goaf, optional TEM sweep | "No rise" taken as safe | Geo/Monitoring | 30–60 s |
| 10 | Data export, backup, drill plan | 90‑day dataset restored from backup once, drill scheduled | Backups never tested | Monitoring/IT/EHS | 60 s |
Tape this small card near the screen. It guides fast pattern reads.
Gentle climb → external inflow grows → check goaf and returns Random jitter → likely bubbles, cable noise, poor seating → run quick hardware check Step jump → valve or pump switch, or short inflow pulse → correlate timestamps Night window rise → external inflow, slow rebound → raise sensitivity at night
Goal: confirm eyes on every hydraulic segment. Bends, T‑junctions, pump‑room return, goaf boundary. One sensor per segment at minimum. Two near high risk for cross‑check.
How to act:
Overlay recent curves from neighboring points. Seek lockstep motion.
Hunt for "decoupled" pairs. Same rain, same shift, yet curves drift apart. A gap points to a blind spot.
Place sensors in stable straight runs near each corner. Keep a short standoff from turbulence.
Pass rule: critical points meet plan or exceed it. Curves agree across neighbors.
Pitfall: a single point at the bend center. Signals get noisy. Trends fade.
Evidence to store: screenshots before and after any tweak.
Goal: a smooth bounce back after power cuts. A real test matters more than a sticker on a battery.
How to act:
Cut power deliberately during a safe window. Record time to full recovery for logging and comms.
Run a UPS load test. Note voltage trend and runtime.
Inspect grounding and panel cleanliness. Dust and moisture cause silent trouble.
Pass rule: full recovery inside three minutes. UPS endurance ≥30 minutes. Battery health ≥80%.
Pitfall: no drills, only hope.
Evidence to store: photo of panel, UPS log, recovery time.
Goal: a clean link for live alerts and curves. Wired runs, radio hops, fiber splices. All in scope.
How to act:
Sample latency and loss on a few key nodes.
Inspect connectors in damp areas.
For radio, read signal bars at sensor level, not only at the gateway.
Pass rule: offline ratio below one percent. Key nodes online for thirty days straight.
Pitfall: ignoring humidity near joints. Corrosion creeps fast.
Evidence to store: link test screenshots, splice photos.
Goal: numbers which reflect reality, time tags which line up across the fleet.
How to act:
Check last calibration date per sensor.
Cross‑compare two nearby sensors under stable flow.
Sync clocks from a common source. Write down skew before and after.
Pass rule: drift inside tolerance. Skew under two seconds.
Pitfall: hardware calibration only, time left adrift.
Evidence to store: calibration certificates, time sync log.
Goal: early warning for real risks, less noise for crews.
How to act:
Use three tiers: tip, warn, alarm.
Tune night sensitivity higher. Quiet lines reveal soft climbs.
Build a contact tree per shift. Include two backups for every role.
Pass rule: fewer false alarms over ninety days. Earlier notice by at least fifteen minutes for known patterns.
Pitfall: one rigid value for all places and hours.
Evidence to store: threshold table, month‑over‑month false alarm count.
Sample threshold table:
| Metric | Tip | Warn | Alarm | Notes |
|---|---|---|---|---|
| Water level change (15 min) | +2 cm | +4 cm | +6 cm | Night window tighter |
| Pressure spike | +5% | +8% | +12% | Cross‑check near high points |
| Flow rise (night baseline) | +3% | +6% | +10% | Compare to dry nights |
Adjust numbers per mine policy and sensor class.
Goal: confirm last‑mile delivery to real humans in real time.
How to act:
Send one test alert to the live on‑duty group.
Ask every recipient to hit "acknowledge."
Record any gap. Fix phone numbers, roles, device tokens.
Pass rule: one hundred percent receipt and confirm in three minutes.
Pitfall: stale contact list after staff rotation.
Evidence to store: delivery receipts, a refreshed list.
Goal: capacity margin before storms. Standby readiness proven, not assumed.
How to act:
Run main pump no‑load for one minute, then load for two.
Swap to standby for one minute. Confirm smooth handover.
Log flow, head, amperage, noise, leaks.
Quick math example:
Design flow: 120 m³/h. Safety factor: 1.2. Target ≥144 m³/h.
Measured: 150 m³/h. Pass.
Head drop vs design ≤10%. If higher, plan an inspection for suction or impeller wear.
Pass rule: active train ≥ design × safety factor. Standby cut‑in under sixty seconds.
Pitfall: only main pump tested. Standby never starts under load.
Evidence to store: gauge photos, short video of swap.
Goal: rapid action in the first five minutes.
Five low‑cost items:
Spare couplings
Quick‑plug power cable
Check valve
Portable lamp
Bold signage for lines and valves
How to act:
Label each item. Fix shelf location codes.
Run a monthly inventory tick‑off.
Train a two‑person team to grab items in a call.
Pass rule: items present, labeled, reachable in five minutes.
Pitfall: a pretty shelf, no labels, no cycle checks.
Evidence to store: shelf photo, inventory sheet.
Goal: higher resolution near old workings. More samples, more clues. Optional geophysics for a quick scan.
How to act:
Double sampling frequency near suspected goaf zones.
Compare night windows across several weeks after first heavy rain.
Consider a fast YCS2000A TEM sweep to paint a conductive "bright line" near edges.
Pass rule: improved curve resolution on risky segments. A drafted response plan in place.
Pitfall: flat line taken as safety during dry weather, then surprise after storms.
Evidence to store: high‑freq plots, a short plan for actions.
Goal: history on hand, backups which restore, a crew which has muscle memory for incidents.
How to act:
Export ninety days of trends and alert logs.
Restore from backup once on a spare machine. Prove it works.
Schedule a half‑day drill: alert → confirm → dispatch → pump swap → review.
Pass rule: verified restore success, a drill report, action items for the next month.
Pitfall: backups never tested, drills always postponed.
Evidence to store: restore log, drill checklist and photos.
A crew in a western panel ran this five‑minute pass on a dry morning. Curves looked calm. Night windows showed a slow climb near one old stope. Two sensors sat on a bend center only. The team moved one sensor five meters upstream into a steady section. A second sensor went downstream for cross‑check. After a rain two days later, the pair revealed a clear step plus a gentle rise. Thresholds kicked a "warn" tier fifteen minutes earlier than last season. Pumps started sooner. Sump stayed below the comfort line. Crews liked the calm.
Paste this on a clipboard. It guides the whole pass.
[Screen check 60–90 s]
Water level → trend ok? spikes?
Pressure → high points stable?
Flow (night) → baseline drift?
[Field check 90–120 s]
Bend A → upstream straight run sensor
Return → pre/post mix points present
Goaf edge → denser grid
[Alert test 30–60 s]
Push test → 100% ack in ≤3 min
[Pump run 90 s]
Main → Standby swap video
Log flow/head
[Log & archive 30–45 s]
Export mini report
Cloud + local copy
Assign names. Keep accountability clear. Rotate weekly.
| Step | Primary role | Backup role | Evidence |
|---|---|---|---|
| Screen scan | Monitoring lead | Shift tech | Screenshot pack |
| Blind‑spot walk | Maintenance | Geo tech | Photos, notes |
| Alert test | Dispatch | EHS | Delivery report |
| Pump run | Mech/Elec | Pump op | Gauge photos, video |
| Log & archive | Monitoring | IT | Export file, checksum |
| Backup restore test | IT | Monitoring | Restore log |
| Drill lead | EHS | Dispatch | Drill checklist |
Keep lockout/tagout for power tests.
Talk to the face crew before any radio or cable move.
Note every anomaly, even a small squeak. Minor noise turns into heat, heat turns into failure.
Share a one‑page digest at the end of the shift. People remember pictures more than long paragraphs.
False alarms drop by 20–40% over ninety days.
Average warning lead time grows by ten to twenty minutes.
Offline minutes per key node stays under 1% of calendar time.
Pump swap time falls under sixty seconds in drills and real calls.
Restore test passes every single month.
□ KJ117(A) curves reviewed → level / pressure / flow
□ Neighbor pairs compared → no decoupling spotted
□ Bends / returns / goaf edge → sensors placed in steady runs
□ Power cut drill done → recovery ≤ 3 min
□ UPS endurance ≥ 30 min documented
□ Comms test → loss < 1% on sample nodes
□ Calibration date reviewed → time skew < 2 s
□ Thresholds tuned → night tier tightened
□ Contact tree refreshed → alert test 100% ack
□ Pump main + standby tested → flow ≥ design × 1.2
□ Go‑bag items labeled → shelf photo taken
□ Sampling near goaf doubled → plan drafted
□ Export 90‑day data → cloud + local copy saved
□ Backup restore verified → log archived
□ Half‑day drill scheduled → calendar invite sent
Five minutes feel tiny. The payoff grows huge once storms roll in. We push a small routine. We buy precious minutes. We harden the signal chain end to end. We make pumps ready. We keep data safe. The mine runs calmer. The team sleeps better.