Solar Pump for a Spring: When Gravity Works Free and When You Pay in Head

The short answer: a solar pump for a spring is the easiest pumping job in this whole series, and the most commonly mis-sized one, for the same reason: the water arrives under its own pressure, at an elevation you did not choose, and the flow rate is on a schedule the sky wrote. Two decisions govern everything. First, measure the spring at the end of the dry season, not after the rains - spring flow follows a predictable decline called the recession curve, and a source that gives 5.4 m3/h in the wet month can honestly deliver under a quarter of that after 150 dry days. Second, do the gravity arithmetic before you buy a pump - a pipe running downhill from a spring box is a pump made of geometry, and 2 to 10 m3/h can move through ordinary PE pipe on nothing but slope. Only when gravity runs out of elevation do you buy a pump, and then a low-head model is usually the right machine, not a deep-well monster.

That second point is the one buyers skip. I get messages describing “a spring on my land” and a pump model in the same sentence, and when I ask for two numbers - how high is the tank above the spring eye, and how far is the pipe run - the honest answer is often that no pump is needed at all, or that a 200 W model does what a 750 W one was about to be bought for.

First, know what kind of spring you have

Springs are not one thing. Where the water comes from and how fast the ground releases it decide whether your supply is steady through the year or a wet-season loan you have to pay back in drought:

The karst row is the expensive lesson. A spring draining open fissures answers rain within hours and abandons you within weeks, because it has almost no storage. A contact spring draining a deep aquifer barely notices a dry month. If your spring is fenced by rock openings and clears to muddy after every storm, treat it as surface water that took a shortcut, and read the contamination test in the protection section before you drink from it.

The recession curve: your spring’s dry season is already scheduled

Every spring’s decline through a rainless stretch follows a curve described by Maillet in 1904, and it is short enough to write in one line:

Q(t) = Q0 x e^(-a x t)

Q0 is the flow when the recession begins, t is days since, and a is the recession coefficient, in per day. Small a means a fat aquifer feeding the spring slowly; large a means a small or open storage that empties fast. Published values for real springs range from under 0.002 for large limestone systems to over 0.05 for small perched ones, which is a 25-fold range - and it is the difference between a source you can pump all year and one that is a puddle by February.

Run the equation for three realistic cases and the trap shows itself:

Every number in that table is the Maillet equation evaluated; nothing is invented. The medium column halves its flow in 69 days and is down to 1.2 m3/h by day 150. The small-storage column loses 95% of its flow in 150 days. If you sized a pump and a village around the day-0 figure of any of these, the hardware is fine and the design is not.

The practical use is a design date. Ask how long your local dry season actually runs - 90 days is common across the Sahel, up to 150-180 in parts of East Africa - and read your spring at the bottom of that column. The flow your pump will actually live with is the flow on the day the rains come back, not the flow on the day you visited the site.

You can reproduce every figure above yourself; we keep the arithmetic in a small script (spring_calc.py) alongside our other published calculators, so nothing here has to be trusted.

Measure it with a bucket and a stopwatch

You do not need a weir or a flow meter to get an honest number:

flow (m3/h) = 3.6 x litres / seconds

Three rules make the measurement worth having. Take three readings and average them; measure at the collection point rather than wherever the water happens to pool; and do the whole exercise again at the end of the dry season, because that second number is the one your design is actually built on. If the spring is too weak to fill a 10 L bucket in under two minutes - under about 0.3 m3/h - the honest answer is that pumping it is not the problem; whether there is enough water to pump is.

The gravity check: the cheapest pump is no pump

Before choosing a model, run the check that costs nothing. Water flowing downhill in a full pipe obeys the same Manning equation that sizes irrigation channels, and inverted it tells you how much slope a pipe needs to pass a given flow:

slope = (v x n / R^(2/3))^2 with v = flow/area, R = D/4 for a full pipe

For ordinary PE pipe (n = 0.013) the results are startlingly low:

Read that table as a decision rule. A spring box built just 2.6 m above the tank pushes 5 m3/h through 300 m of DN63 with no pump, no panels and no controller - the head comes from geometry you already own. The same 5 m3/h through the same pipe driven by a pump needs 2.55 m of head, which any small model supplies easily, but the pump version also needs the array, the controller and a reason to exist.

The full-pipe capacities are equally useful in the other direction:

Two honest caveats. The pipe must actually run full for these numbers to hold - a pipe with a high point traps air and the capacity collapses, which is why gravity lines are laid with continuous fall and no humps. And gravity is rigid: if the tank site is fixed and the fall is 1 m, no amount of wishing turns that into 2.6 m. When the arithmetic says the fall is not enough, that is the moment a pump enters the story - and the sizing from here on is unusually kind.

Sizing the pump: low head is the normal case, not the compromise

A spring’s head budget is the lift from the water level in the collection box to the tank, plus friction in the delivery pipe - there is no drawdown, because the spring refills as you take. That single fact changes the model ladder completely. Where a borehole at the same site might demand 80-160 m of head, a spring job commonly lands between 5 and 45 m, and the catalogue has its friendliest residents there:

Two ways to read this table. First, flow varies 40-fold across the ladder while head varies barely 2-fold - spring pumping is a low-head discipline, and the design effort goes into flow matching, not head conquering. Second, compare the last two rows: the same 1500 W is sold as either 20 m3/h at 48 m or 11 m3/h at 60 m, and neither is better. They are different machines for different duty points, which is the same lesson the yield-test guide teaches from the borehole side.

A worked example. Suppose the recession arithmetic says the spring delivers 2.2 m3/h at the bottom of a 120-day dry season, the tank stands 28 m above the collection box, and the delivery run is 180 m of DN50:

• Static lift: 28 m
• Pipe friction at 2.2 m3/h in DN50: about 2 m over 180 m (from the same Manning arithmetic in reverse)
• Total head: about 30 m
• Daily harvest: 2.2 m3/h x 5.5 effective sun hours = about 12 m3/day

The fit is 4DSC9.5-50-110-750: at 30 m of head it is nowhere near its 50 m limit, it supplies far more than the spring’s dry-season flow, and because the pump tracks the solar input, it simply throttles back to what the spring gives. A 2DPC1.5-35-24-200 would also clear 30 m on paper - but it cannot pass 2.2 m3/h at that head, so the spring’s water would be left in the ground. The flow the spring offers on its worst day, at your actual head, is the sizing point - not the maximum flow on any label.

One electrical note that matters more here than on wells: 24 V and 48 V models are fine at these low powers, and their controllers are simpler and cheaper. The 110 V platform earns its complexity above roughly 750 W.

Storage: what the tank does when the spring and the sun disagree

On a spring, storage answers a different question than on a borehole. The well question is “how do I spread 4 hours of pumping across 24 hours of use”; the spring question is “what happens on the day the spring dips and the sky is cloudy at the same time” - the coincidence that dry seasons serve up regularly.

The conservative pattern is to store 2-3 days of demand rather than one:

The recession table tells you when this matters. On the medium-aquifer spring, flow falls below half its wet-season value after 69 days; if your storage covers two days, the system rides through the cloudy spells of early dry season without a thought. Storage is also what lets the pump take everything the spring offers on its best days - the surplus fills the tank and pays for the lean ones. That is the same “store water, not electricity” logic from the battery-versus-tank guide, applied to a source that fluctuates by season instead of by hour.

Protect the spring, or the pump just moves dirty water faster

A pump on an unprotected spring is an efficiency upgrade for contamination. The protection standards used across rural water supply programmes are cheap, physical, and mostly dug or fenced rather than bought:

That last row is the honest test. A genuine groundwater spring stays clear through storms; one that clouds up after rain is delivering surface water through a short hole in the ground, and it needs the box and ditch more than any other item on the list - or treatment, if the shortcut cannot be sealed. These figures are the standing guidance of WHO-based source-protection programmes (CAWST, WEDC technical briefs, and national rural water manuals all converge on the same numbers).

When a spring and a pump do not mix

Three cases where I tell buyers not to pump the spring:

1. The recession says the spring dies before the season ends. If the 180-day column of your curve is near zero and your demand window includes month five, no pump helps; the water is not there. The honest alternatives are a deeper source (a borehole into the same aquifer), or reducing demand. Our borehole yield-test guide covers that path.
2. The spring eye is inaccessible or legally shared. A collection box needs to be built at the eye; if the source serves a village upstream of you, the box and the fence need the community’s agreement first, and in many regions the water rights need checking before any concrete is mixed.
3. Gravity already does the job. If the table above says your fall covers your flow through your pipe run, spending money on a pump buys you nothing but maintenance. Put the budget into tank volume instead - it is the one upgrade that never needs a controller.

And one case where people over-worry: pumping a spring does not harm it. The spring flows whether you collect the water or not; a pump that takes less than the recession-curve minimum simply leaves the surplus to spill over the box lip. The failure mode to design against is not “pumping hurts the spring”, it is “pumping claims a flow the dry season will not honour.”

The decision, in order

1. Measure the spring at the end of the dry season: bucket, stopwatch, three readings.
2. Sketch its recession with two measurements across the season and estimate where the bottom is.
3. Run the gravity check: fall available versus fall needed for your flow and pipe run. If gravity wins, build the tank bigger.
4. If a pump is needed, total head = lift from box to tank + friction. Pick the model whose flow at that head exceeds the spring’s worst-day flow.
5. Store 2-3 days of demand. Protect the spring before the first panel goes up.

Done in that order, a spring system is the most durable thing in rural water: no fuel, no drawdown, no water table racing you downward - just a source the ground refills, a pump that works when the sun does, and a tank that carries the difference. If you send me your three bucket readings, the height of your tank above the eye and your pipe run, I will return a recession estimate, a gravity verdict, and a model and panel count that matches the spring’s worst month, not its best.

Building on a spring? Send me your bucket-timing readings, the tank height above the spring eye and your pipe run on WhatsApp and I will tell you whether gravity alone covers it, or come back with a model, a panel count and a tank size sized to the recession curve. For borehole sources on the same land, start with the yield-test guide, and for the storage side read battery versus tank.