Pump Reliability & Predictive Maintenance (PlusSmart) · 9 min read · Updated November 2026
A Pump Health Check from the Data You Already Have: KPIs, Ranking, Bad Actors
Your historian has logged every pump's pressure, flow, speed and power for years. Your CMMS has every repair. Nobody has read them together. A Pump Health Check does exactly that — pump engineers compute five KPIs per machine, rank the fleet from good to bad, and hand back a list of actions — before a single sensor is bought.
Only a small fraction of the data a plant collects is ever analysed, and pump data is the least read of all: it sits in the DCS, SCADA, drives, protection relays, the CMMS and — in older plants — in handwritten shift logs. PlusSmart Analytics is an engineer-led service that reads it. Not machine learning that needs a hundred failures to train; pump engineers applying pump physics (curves, BEP, affinity laws, NPSH) to the data from the first day, explaining causes rather than flagging anomalies. The first deliverable is the Pump Health Check, and it is the right starting point for any reliability programme because it tells you where to put the next dollar.
What data counts
| Source | Typical content | What it reveals |
|---|---|---|
| Variable speed drives | Speed, current, torque, power, fault log | Operating point, abnormal load, dry running, energy |
| DCS / SCADA / historian | Suction and discharge pressure, flow, sump level, run status | Actual head, efficiency, starts, run hours, time off-BEP |
| Protection relays, power meters | Three-phase current, energy, power factor | Phase unbalance, overload, specific energy |
| CMMS | Work orders, parts, failure dates | MTBF, MTTR, cost per machine, repeat failure modes |
| Shift logs, maintenance diaries | Pressures, temperatures, noises, oddities | Trends from before digitalisation; context for events |
| Pump datasheets and curves | BEP, NPSHr, impeller trim | Reference for everything above |
Handwritten logs are data: many plants start there. Nothing has to leave the site that you do not choose to share; exports of a few pumps' trends are enough for a first pass.
The five KPIs
- MTBF and MTTR per pump — who fails most and how long each repair takes. The distribution is always skewed: ten pumps cause most of the cost.
- Flow as a percentage of BEP, time-weighted — pumps that spend most hours outside 70–120 % of BEP are the future seal and bearing failures, whatever the maintenance schedule says.
- Specific energy (kWh/m³) and wire-to-water efficiency — compare pumps on the same duty; a worn or mis-sized pump stands out as the energy-hungry one.
- Starts per hour — frequent starts overheat motors and produce water hammer; usually a control-logic problem, cheap to fix.
- Life-cycle cost — energy is typically the largest component of pump ownership cost, far above purchase price; the KPI puts it next to repair cost per machine.
What comes back
Asset ranking. Every pump and motor ranked from good to bad with its KPIs and deviations from your own procedures highlighted. Bad actors early. The pumps that still "look fine" but spend their hours in the wrong place on the curve — identified before the third failure, not after. Knock-on effects. The process change in area A two months ago that raised failure risk in area B; maintenance rarely knows, the data shows it. Forecast and budget. Where each machine is in its life, remaining life, risk and cost — for next year's maintenance and staffing budget. An immediate action list. Typically alignment, throttled discharge valves to open, control setpoints to change, two or three pumps to protect with the drive — stabilise operation first, then schedule the maintenance.
Service packages
| Package | Scope | Deliverable |
|---|---|---|
| Pump Health Check | Review of existing data, site survey, ranking | Ranked fleet, immediate action list, pilot proposal |
| Custom analytics and integration | KPIs and reports per asset and team; cloud or customer data lake; sensor and CMMS integration | Data anywhere, automatic work orders, continuous improvement |
| Pump expertise for your team | Commissioning VFD pump systems, optimising and rewriting PLC pump control, ISO 50001 energy management | Pump systems running as designed; data-driven budgets |
Why not just buy software?
AI predictive-maintenance platforms learn from failures, and critical machines are rarely allowed to fail often enough to teach them; they also do not know that a pump at 55 % of BEP flow is in trouble unless someone tells them. Pump engineers do, from the first week of data, and they explain the cause in terms the planner can act on. The analytics platform is the tool; the engineer is the product. Once the fleet is ranked, sensors (i-ALERT, ElectroSine, Poseidon) go where the ranking says, and ReliaPump control goes on the stations whose data shows dead-heading — the Health Check is how the rest of the programme is sized.
Need engineering help with this?
Our engineers apply this to real projects every week — selection, specification, commissioning. Send your question or a part number.
Quick answers
Does our data have to leave the plant?
You decide the scope. A first pass can use trend exports for a handful of pumps; later, a cloud platform or a customer-owned data lake are both supported.
How long does a Pump Health Check take?
Typically two to four weeks from receiving the data exports and a site visit, depending on fleet size and data quality.
What if we have no historian?
Drive fault logs, CMMS records and shift logs are enough to rank by failures and cost; the KPIs that need process data are added as sensors or SCADA exports become available.