A pump station that runs on a pair of float switches and hope will fail at 2 AM, during the storm, with a full sump and a drowned basement. That is not pessimism — it is the standard failure mode of unautomated pumping. The fix fits in a 35mm DIN-rail module and costs less than one emergency call-out.
The classic pump-control wishlist, all doable on a compact CPU with a handful of function blocks: pump alternation, dry-run protection, level alarms, run-time logging, and a remote alarm channel that does not depend on someone walking past the building.
Alternation: The Feature That Doubles Pump Life
Two pumps, one always leads, one always lags — that is how pumps die young. The lead pump accumulates all the starts, wears out, and the lag pump stays pristine untill the day it is suddenly needed and fails from disuse. Alternation logic swaps the lead every cycle or by run time. Both pumps share the load, and the station keeps working when one pump fails, becuase the other has been kept in practice.
That is a handful of blocks: run-time counters, a selector, two output coils. The payoff is measured in pump-replacement invoices you never receive.
Dry-Run Protection and Level Logic
Running a pump dry for minutes ruins the mechanical seal. A low-level sensor or a level-transmitter comparison trips the pump off before the damage — and, crucially, resets automatically when the level recovers, with a restart delay to keep the pump from short-cycling. The analog input reads a 4-20mA level transmitter; the comparator sets the thresholds; the program handles the rest.
On a 4-20mA loop you also get free wire-break detection — a current below 4mA is a fault, not a level. Wire the alarm logic for it and the station reports its own sensor failures.
The Remote Alarm Question
Nobody sits at the pump station. The alarm must travel. Pick the channel that fits the site:
- Ethernet CPU plus the free EasySCADA package — full trend and history on a PC
- Modbus TCP to a cloud or SCADA — standard, protocol-level
- SMS on a GSM-capable model — the station texts the duty engineer
- Web server — the browser page shows live status to anyone with the URL
Each option is a configuration step, not a hardware redesign. The same CPU covers the site today and the network expansion next year.
Real Case: Lift Station Retrofit
A small lift station with two pumps ran on a relay panel that had seen better decades. The retrofit: one compact CPU, a 4-20mA level transmitter, a GSM alarm module on the serial side, alternation and dry-run logic in about fifty blocks. Commissioning took a morning. The first real storm hit two weeks later — the station cycled through both pumps, sent one SMS, and nobody had to visit.
The contractor's report used the word "boring." I have never heard higher praise in this industry.
For a remote site with no network in sight, the same logic fits in one compact unit with the alarm channel built in — see the PR-26DC-DAI-RT-4G pump-group build with 4G SMS alarms for the concrete wiring, program and test-bench numbers.
Design Notes Before You Build
Give each pump its own overcurrent protection and a manual mode — the panel must run without the PLC when required. Keep the float-switch and transmitter inputs on separate channels so a failed sensor does not take down the whole station. And log the run hours; the alternation logic needs them, and so does the maintenance schedule.
The Maintenance Plan the Data Writes
The old way: service the pumps on a calendar. The better way: let the controller's run-time counters and start logs write the schedule. A pump that logs 400 starts a month gets a different service plan than one logging 40 — and the controller knows the difference while the calendar does not. The data is already in the flags; the discipline is using it.
Expose the counters on the operator screen, the SCADA, or the SMS reports. When the run-time approaches the service threshold, the alarm chain fires — the same chain that already notifies the duty engineer. The maintenance visit happens because the data said so, not because the pump failed. That is the difference between planned work and emergency work, and the price difference is an order of magnitude.
The alternation logic needs the counters anyway — the lead pump is chosen by run time, keeping the wear balanced — so the data collection is free. The maintenance plan is the same data wearing a second hat, and the hat costs one alarm block.
Commissioning the Station in One Morning
The retrofit story from earlier is repeatable, and the recipe is short. Step one: wire the level transmitter, the pump outputs, and the float-switch backup inputs. Step two: load the program — alternation, dry-run, alarm, restart delay — from the library. Step three: test every fault path on purpose: low level, high level, sensor failure, pump fail-to-start. Step four: configure the alarm channel — SMS, Ethernet, or both. Step five: the log is running.
Five steps, one morning, and the station is done — because the platform blocks are standard and the faults are simulated, not discovered. The commissioning checklist exists to make step three brutal: the station that fails every test in the workshop is the station that works at 2 AM in a storm.
That is the honest measure of a pump retrofit: not how fast it installs, but how it behaves when the storm arrives. Test for the storm, and the storm becomes a footnote.
Pump control is not glamorous. It is just the most profitable application in the whole micro-PLC catalog, because the alternative is measured in flooded basements and emergency invoices.
Related: water pump group control with the PR-26DC-DAI-RT-4G · EasySCADA setup
