The Second Pump: Duty, Standby and Redundancy Without Buying Twice

The short answer: the second pump is not something you buy — it is something you design. Ask a farmer what happens when the solar pump stops and the honest answer is a shrug. Ask what happens when it stops on a heat week with a nursery on drip, and you get a price: wilted seedlings, stressed cows, a pond gasping by evening, a village walking back to the old source. That price — not the catalogue — decides whether you need a second machine, a bigger tank, a spare-parts kit, or just a monthly fifteen-minute test. This guide turns redundancy from an afterthought into a short calculation, and then maps the layers onto the model ladder, every model verified against the live product database.

Price the outage, not the pump

Redundancy is insurance, and insurance is priced against the loss it prevents. The loss is almost never the pump itself — at these prices the pump is the cheapest casualty of its own failure. The loss is the system around it:

One large-scale review of mining operations puts unplanned downtime at hundreds of thousands of dollars per day — which is why dewatering codes treat standby pumps as safety equipment, not options, and why our mine-drainage guide reads its redundancy rules as a sizing checklist. A village sits at the opposite pole: little cash loss, but a health event and a trust failure that outlasts any repair. Most farms sit in between — real loss, not instant loss — which is exactly the regime where layered protection beats a twin machine.

Breakdowns are normal: thirty years of data

If breakdowns were rare, redundancy would be paranoia. The rural water sector has measured them for decades, and the numbers are stubbornly stable:

• The RWSN synthesis across 20 sub-Saharan African countries put average handpump non-functionality at roughly one in three — about 36 percent, with country rates from 10 to 65 percent (RWSN, 2009).
• The most recent comprehensive analysis (Foster et al., 2019, using 2015 data from 38 countries) improves the picture only slightly: about one in four handpumps down at any moment, from 11 percent in Burkina Faso to 30 percent in Togo and Côte d’Ivoire.
• National studies since have run worse, not better: over 38 percent of improved water points non-functional in Nigeria, 29 percent of Afridev handpump boreholes in Malawi, 29 percent of handpumps in Tanzania.
• Response time, not spare parts, dominates the downtime: monitoring research in Kenya found handpumps sitting broken for 30 days or more per breakdown, while professional maintenance services push uptime above 90 percent by shortening response — not by preventing every fault.

Read that as a solar-pump owner: a breakdown is a when, not an if, and the variable you control is not the failure rate but the recovery time. Every layer below is a way to buy back time.

The four outage classes — and the layer that closes each

Every pump outage belongs to one of four classes, and each class has a cheapest cure:

The table is the whole strategy. Classes A and B close for wiring and a few spare parts; class C closes with a kit and a phone number; only class D — the rare, total losses — justifies a second machine. Buying a twin to cover class A is the most common over-buy in solar pumping, and skipping the class A wire because you own a twin is the mirror-image mistake.

Layer 1: storage is standby you already paid for

A tank covers every outage that ends faster than the tank drains, and it never needs a service visit. The arithmetic is demand-side: two days of essential demand for a household or a herd, one peak day for an irrigation block. A dairy drawing 35 m³ a day needs at least 70 m³ of usable water for two days without inflow; 20–30 m³ usable provides only about 13.7–20.6 hours at constant average demand; a village drawing 120 m³ wants a 300–400 m³ reservoir, which is also the buffer that lets one pump serve a scheme designed for growth. Water in a tank does not care what broke — cloudy week, tripped float, a controller on a truck — it simply carries the system across the gap. The sizing rules live in our storage-tank guide; the redundancy view adds one habit: lift the float valve by hand at every service visit, because a tank behind a stuck valve is scenery, not standby.

Layer 2: the A/D input — one machine, two fuels

The catalogue already contains the cheapest standby in solar pumping, and it hides in the model name. From the 750 W tier upward, -A/D models accept DC from the array and AC from grid or generator — two independent energy paths into one hydraulic set. The mine regulations phrase the pattern perfectly: the solar array is the duty source, the AC input is the standby. Concretely: a 3DSC6-60-110-750-A/D (750 W, 110 V) runs its borehole all day on solar and, when the weather turns, keeps the same hydraulics alive from a generator; a 6DSC65-85-380/520-7500-A/D (7,500 W; AC380/DC520) does the same at scheme scale. The A/D input closes classes A and B-power in one wire — but not class C or D: AC power is no help to an absent impeller. That residual is what the next layers are for. For the full dual-source logic, see our DC vs AC/DC guide; the operational rule is one line — run the system on AC for fifteen minutes every month and log it, because an untested input is a hypothesis, not a standby.

Layer 3: the spare-parts kit that matches this catalogue

The kit is sized by one question: what fails in the first week of a breakdown? For this catalogue the answer is short — the float switch and dry-run probe, the seal set, a spare impeller matched to your model, the controller’s fuses and connectors, and the gaskets for the pump’s outlet joints. Store it in the controller box where it cannot walk away, and add the two things no kit ships with: the puller for your well head, and the phone number of the technician who can use it. A kit converts class C from a week of downtime into an afternoon — provided someone within a day’s travel can actually open the pump. Above roughly a kilowatt, pull time becomes the bottleneck, which is the honest boundary of what a kit alone can promise.

Layer 4: the real second pump — cold standby done right

When the loss calculation says the second machine pays, buy it properly:

1. Match the hydraulics, not the model number. The standby should meet the same duty — flow at head — so a 9 m³/h duty at 90 m is backed by another 90 m-class machine, not by whatever was on discount. Identical is simplest: same model, same controller, same spares.
2. Cold storage, not hot standby. Store the complete pumpset — pump end, controller, float, cable — dry and indoors, plugs taped. A hot standby spinning against a closed valve wastes bearing life every day waiting for a failure that may never come.
3. Exercise it monthly. Ten minutes against an open discharge from a temporary array or AC source. The test log is the difference between a standby and inventory.
4. Decide the switchover before the crisis. Union fittings on both riser pipes, or a bypass with valves, so the changeover is five minutes of wrench work — not an improvised plumbing project in the dark.
5. Alternate the duty on a schedule if you prefer two running machines: swap duty and standby every month or quarter, and both ages evenly while each remains a genuine backup. This is the water-utility pattern, and it doubles as the monthly test.

The model ladder, read as a redundancy ladder

The same fourteen-model ladder that runs through every guide on this site, read here as outage insurance:

* 18,500 W per the specification table; the model name carries the 11,000 figure. All other figures verbatim from the catalogue.

Read the ladder against the outage-class table: below about 750 W, a kit and a tank close most of the risk at kit prices. From 1.1 to 2.2 kW the A/D input arrives and class A closes for the price of a wire. From 7.5 kW the scheme is worth monitoring and a contract response — see our remote-monitoring guide — and at the top tier the duty/standby logic of the mining codes applies, so the twin machine is specified by policy, not debated per outage.

Two complete designs

Hypothetical dairy-farm design with a 1,500 W pump candidate. Measured dynamic water depth is 30 m below the wellhead ground; the open tank’s delivery water level is 5 m above the same reference. Suppose a separate calculation of all pipe and fitting losses at the target flow supplies 4 m; required pressure head at the open tank is 0 m. TDH = 30 + 5 + 4 + 0 = 39 m. Dynamic level already includes drawdown, so do not add it again. In general, TDH = static water depth + drawdown + delivery elevation + calculated pipe/fitting losses + required pressure head. Drilled depth is an installation constraint, and losses need actual internal diameter, total length, material, flow and fittings rather than a fixed distance ratio. Demand for drinking and washing is assumed to be 35 m³/day; over 5.5 equivalent pumping hours the target is about 6.4 m³/h at 39 m. The existing 4DSC9.5-90-110-1500 (1,500 W, 110 V) remains a candidate pending a measured-curve check. Six 550 W modules give 3,300 W against the 1,950 W power floor; voltage and current compatibility also need checking.

A full tank with 20 m³ of usable water supplies 20 ÷ 35 × 24 ≈ 13.7 hours, assuming constant average demand and no inflow. Less water at the start means less cover; nominal tank volume may exceed usable volume. Two days at 35 m³/day require at least 70 m³ usable storage. Keeping only 20 m³ for two days requires either a separately validated rationing plan of at most 10 m³/day, or, at unchanged demand, reliable pumping that supplies at least 50 m³ during those two days and arrives before the tank empties. Panel cleaning and spare connectors do not guarantee cloudy-week pumping. Keep the spare-parts kit (float switch, dry-run probe, seals, fuses and impeller), but set the repair response against the actual remaining tank autonomy; a low tank cannot promise two days of cover.

A growth-stage water scheme for 3,000 people — where the cold standby earns its place. Demand: 3,000 people at 40 L each, plus a livestock trough — about 130 m³/day, a duty near 24 m³/h at a 35 m head. The 6DSC65-85-380/520-7500-A/D (7,500 W; AC380/DC520) runs it mid-curve with room to grow; twenty 550 W modules give 11,000 W against the 9,750 W floor. Redundancy in four layers: a 400 m³ reservoir — three days of essential demand; the A/D input wired to the scheme’s generator and proven by a logged monthly AC run; telemetry on flow and tank level, because a scheme this size deserves to know it is down before the first complaint; and a cold-standby complete pumpset — a second 6DSC65-class machine stored dry in the pumphouse, shaft turned monthly, commissioned within a day. The policy logic is the mine code’s: one lost day returns 3,000 people to untreated sources, which costs more than the standby costs to keep. Four layers, one machine running, and nobody carries water twice.

What does not count as redundancy

1. An untested A/D input. Never seen AC, never carried water on it — a hypothesis, not a standby. The monthly logged run is the standby.
2. A bare pump end in a warehouse. A wet end without its controller is half a machine. Store the pair, or budget the controller in the standby line.
3. A tank behind a seized float valve. Lift the float by hand at every service visit; storage only insures what it can actually deliver.
4. A spare impeller for a pump you cannot pull. Without a puller or a crew, the impeller is inventory, not insurance. Budget the pull first.
5. Insurance bought twice at every tier. A twin and every kit and every layer is over-insurance. Match the layer to the class table, stop there, and spend the difference on water.