Battery Solution for Water Treatment Pumping Stations
Pumping stations are the least photographed asset in a water utility and the one that punishes a power outage fastest. A treatment plant can coast for twenty minutes while operators switch sources. A wet well cannot. Once the level reaches the high-high float, the remaining options are a sanitary sewer overflow, a flooded basement, or a boil-water notice. I have spent a decade specifying DC-coupled backup for municipal and industrial sites, and I have never seen a pumping station where the sizing conversation ended at the nameplate. This article walks through how I engineer a battery solution for water treatment pumping stations, from hydraulic energy and starting kVA to flood elevation, hazardous area classification, and the maintenance calendar.
What A Pumping Station Actually Asks From A Battery
A station does not draw a steady current. It draws almost nothing, then a great deal, in bursts, so size each load block on its own terms.
The duty cycle is lumpy, not flat
Process pumping dominates. A 22 kW (30 hp) end-suction pump running at 60 percent duty over an eight-hour outage consumes roughly 105 kWh before any losses. Chemical feed pumps look trivial at 0.2 to 0.5 kW, but they run continuously and they decide whether the effluent stays compliant, so they belong on the critical bus with no shedding allowed. Mixers, flocculators and blowers sit in between, typically 3 to 15 kW each, and they are the first blocks I drop when state of charge falls below 40 percent.
Controls and telemetry never sleep
PLC racks, level transmitters, radios and SCADA gateways draw 200 to 500 W and must ride through the entire event without a single dropout. I put them on a dedicated feed with their own small UPS inside the battery enclosure, so that a pump start never browns out the telemetry. That one decision has saved more sites than any increase in battery capacity: an operator who can see the wet well level can make good decisions, and one staring at a blank screen cannot.
Sizing From Hydraulic Energy, Not From Nameplate Horsepower
The fastest way to overbuild is to multiply motor nameplate kilowatts by outage hours. Hydraulic energy is what you are actually buying.
Take a lift station moving 0.05 cubic metres per second (about 790 gallons per minute) against 30 metres of total dynamic head. Hydraulic power is 1000 x 9.81 x 0.05 x 30, which is 14.7 kW. Divide that by a pump efficiency of 0.75, a motor efficiency of 0.92 and a VFD efficiency of 0.97, and the electrical draw at the meter is 22.0 kW. That is the number that belongs in your energy budget.
Now apply duty. With the lead pump at 60 percent duty over an eight-hour outage, the energy is 22.0 x 8 x 0.60, or 105.6 kWh at the AC terminals. Station house load of 1.5 kW for eight hours adds 12 kWh. The DC bus has to deliver 117.6 kWh.
Convert to nameplate with three factors I never skip: inverter and DC conversion losses of 0.92, usable depth of discharge of 0.90 for an LFP pack held between 10 and 100 percent state of charge, and end-of-life derating of 0.80. The product is 0.662. Dividing 117.6 by 0.662 gives 178 kWh, so I would quote 180 kWh as three 60 kWh lithium battery packs with independent BMS and fusing. The end-of-life factor is the one clients want to delete, and it decides whether the station still meets its autonomy target in year twelve.
Power rating is a separate question from energy. Continuous inverter rating must cover the settled load with margin, while the surge rating must cover a pump start lasting two to three seconds. That is the number a custom battery solution has to be built around.
Starting Surge Decides The Inverter, Not The Motor Rating
A 22 kW, 400 V three-phase motor draws about 40 A at full load, and the starting method changes the inverter requirement by a factor of five. I never specify an inverter before I know how the pump starts.
Direct-on-line, soft starter, or VFD
- Direct-on-line. Inrush runs 6 to 7 times full load current, so 240 to 280 A for 0.5 to 3 s. At the low power factor typical of a locked rotor, roughly 0.3 to 0.4, apparent power is 1.732 x 400 x 260, which is 180 kVA. That is what the inverter has to source, and it is nearly eight times the motor rating.
- Soft starter. Two to three times full load current for 5 to 15 s, so 80 to 120 A and 55 to 83 kVA, but held for much longer.
- VFD. About 1.1 to 1.5 times full load current ramped over 10 to 30 s, so 44 to 60 A and 30 to 42 kVA. A VFD is the single most effective way to shrink the power electronics, and on retrofits it usually pays for itself.
Manufacturers quote surge over 100 milliseconds, useless for a pump start lasting one and a half seconds. Ask for three-second and ten-second ratings in writing, then hold 30 to 40 percent margin above the computed kVA.
Autonomy: How Many Hours And Who Decides
The autonomy target is a risk question before it is an engineering one. In my experience the defensible numbers are 4 hours for a lift station serving a small collection area, 8 hours for a station feeding a treatment plant, and 24 to 72 hours where an overflow reaches a sensitive water body or where a consent decree is already in force. Write the target down and get it signed before you size anything.
Two multipliers matter. First, infiltration and inflow during a storm can double or triple inflow, so design against the wet weather curve rather than the dry weather average. Second, the battery has to be recharged. If the outage ends after eight hours, the recharge window is six to ten hours before the next event, so the charger is sized near 125 percent of average discharge power. I size the charger at 25 kW for the 180 kWh example, or 40 kW where repeat events within a single storm are common.
Where a generator already exists, I do not remove it. It stays as the long-duration source while the battery carries the first four hours and every short interruption. That combination cuts generator run hours by 60 to 70 percent, reduces fuel consumption and noise, and eliminates the 10 to 30 second window in which the wet well keeps rising while the generator cranks.
Site Conditions That Kill Batteries Faster Than Cycling
Most battery failures I inspect in the water sector have nothing to do with cell chemistry. They are enclosure and environment failures, and they follow predictable patterns.
Wet well atmosphere and corrosion
Hydrogen sulfide is the enemy. At 10 to 50 ppm it attacks copper, silver and tin plating, and the headspace of a sewage wet well is where that happens. Use tinned copper busbars, sealed contactors, conformal coated circuit boards, and an enclosure rated at least IP54 in a dry room or IP65 / NEMA 4X where washdown is routine. Specify stainless 316 or marine grade aluminium for anything within splash distance. I also place a sacrificial corrosion coupon beside the enclosure and inspect it at every service visit, which is the cheapest early warning available.
Flood elevation is not optional
Pumping stations sit in flood plains by definition, because that is where water collects. Set the battery base at least 300 mm above the finished floor and above the documented 100-year flood elevation. An inundated pack is a total loss, and usually an insurance argument nobody wins.
Washdown and temperature
If operators hose down the room, the enclosure needs IP65 and a drainage path that prevents water pooling against a gland plate. LFP must not be charged below 0 degrees Celsius, so an unheated room in a cold climate needs a 200 to 500 W thermostatically controlled heater and a BMS that blocks charge until the pack is above freezing. Above 40 degrees Celsius ambient in summer, derate continuous current or add forced ventilation.
Codes, Standards And The Permit Path
The paperwork determines the schedule as much as the hardware does. This is the stack I assemble on every project, before the first equipment quote.
- NFPA 820 for fire protection in wastewater collection and treatment facilities, which drives hazardous area classification. Wet wells and their headspaces are frequently classified, so the battery must sit in an unclassified room or inside a purged and pressurised enclosure.
- NFPA 855 for energy storage installation, NFPA 70 Article 706 for the storage system itself, UL 9540 for the system, and UL 1973 / UL 9540A for modules and thermal runaway propagation testing.
- NEC Articles 700, 701 and 702 to establish whether the station is legally required standby or optional standby. That classification changes both the allowable transfer time and the inspection frequency.
- IEC 62619 and IEC 62133-2 for cell and pack safety, plus UN 38.3 for transport, which also sets the 30 percent state of charge shipping rule.
- IEEE 446 for emergency and standby power practice, still the clearest written guidance on load block sequencing for motor-heavy sites.
Where the room cannot be made unclassified, the alternative is a purged enclosure with pressurisation monitoring interlocked to the BMS. It costs more and adds a permanent maintenance item, so I prefer to move the battery to a separate room when the layout allows.
Commissioning, Testing And The Maintenance Calendar
Commissioning is the phase everyone compresses and the phase that generates callbacks. My acceptance sequence has five tests.
- A step load test that injects the largest single motor at the design state of charge, measured with a recording power analyser across the full three seconds.
- A worst-case sequence test that reproduces the wettest day you have data for, with pumps cycling lead and lag, while logging module temperatures. Any spread above 5 degrees Celsius between modules needs investigation before handover.
- An autonomy discharge at the design duty profile down to the low state of charge cutoff, then a timed recharge to prove the charger returns the pack to full inside the stated window.
- Failure mode drills: open the DC breaker with the station live and confirm the transfer, then simulate a lost BMS communication link. Both should be survivable without operator intervention.
- Documentation: as-built single line, equipment schedule, cell-level test sheets, enclosure IP test report, torque log, and a BMS event log export with clock alignment confirmed.
I specify a monthly automatic self test, an annual capacity test, annual infrared thermography at every bolted joint, torque checks on DC connections every two years, and coulomb counter recalibration against a full charge every twelve months. A good LFP pack delivers 3,000 to 6,000 cycles, which at one cycle per day is eight to sixteen years, so budget replacement at 80 percent of original capacity as a planned capital item rather than a surprise.
Five Mistakes I Keep Finding On Real Sites
- Sizing from pump nameplate horsepower with no duty cycle and no efficiency chain. This overbuilds run time and underbuilds power.
- Accepting a surge rating measured over 100 milliseconds. Pump starts last seconds.
- Putting the battery in a classified space because that is where the spare wall happened to be. The retrofit costs more than the correct enclosure.
- Forgetting the recharge path. A battery that carries eight hours but needs thirty to recover is a one-shot asset.
- Leaving SCADA on the pump bus, so a pump start browns out the telemetry and the operator goes blind exactly when visibility matters most.
Frequently Asked Questions
How long can a battery solution for water treatment pumping stations run during an outage?
It depends on the duty cycle you design for. A typical municipal lift station with one 22 kW lead pump at 60 percent duty needs roughly 118 kWh at the DC bus for eight hours, which becomes about 180 kWh of nameplate lithium battery capacity once conversion losses, usable depth of discharge and end-of-life derating are applied. Stations feeding a treatment plant are usually designed for eight hours, while remote stations discharging near sensitive waters are frequently specified at 24 to 72 hours.
Can a battery solution replace the standby generator at a pumping station?
It can for most outages, but I rarely recommend full removal. A battery covers the first four hours, every short interruption, and the critical 10 to 30 second window before a generator would normally pick up load. Keeping the generator as the long-duration source cuts run hours by 60 to 70 percent while preserving an unlimited-duration fallback. That hybrid is usually the cheapest resilient configuration.
Do I need to derate a battery for cold weather at a pumping station?
Yes, for charging if nothing else. LFP cells must not be charged below 0 degrees Celsius because lithium plating becomes irreversible and capacity loss accelerates. Specify a thermostatically controlled 200 to 500 W heater, and confirm the BMS blocks charge until cell temperature rises above freezing. Discharge performance is far less affected.
What battery chemistry is best for water and wastewater pumping stations?
Lithium iron phosphate is the default for stationary pumping duty. It offers 3,000 to 6,000 cycles, excellent thermal stability, no cobalt, and a flat voltage curve that suits inverter DC buses. Sodium-ion suits unheated remote sites because it retains 85 to 92 percent of capacity at minus 20 degrees Celsius and can be transported and stored at zero volts, which simplifies logistics for remote sites.
Does a battery solution for water treatment pumping stations need a hazardous area rating?
Only the location does, not the chemistry. NFPA 820 governs classification in wastewater facilities, and wet well headspaces are frequently classified because of methane and hydrogen sulfide. The cleanest answer is to place the battery in an unclassified room; if that is impossible, a purged and pressurised enclosure with monitoring interlocked to the BMS is the accepted alternative, though it adds cost and a maintenance obligation.
How do I know the battery is still healthy after five years?
Three signals, checked on a schedule. First, an annual capacity test that measures delivered kilowatt-hours against the original rating; below 80 percent means the autonomy target is no longer met. Second, module temperature spread under load, which flags a weak cell long before capacity drops. Third, coulomb counter agreement with a full charge recalibration, confirmed every twelve months. A BMS event log export alongside those three checks tells you more than any single state of health number.
A pumping station battery is not a commodity purchase. It is a hydraulic problem, an electrical problem, a corrosion problem and a code problem wearing one enclosure. Size it from the pump curve and the wet weather inflow record, specify the surge rating the starting method actually demands, keep it dry and above the flood line, and put the recharge path and the test calendar in the contract before the first truck arrives.
Further Reading
References
