Designing a Solar Water Pumping System: Head, Flow and Array Sizing

Most failed solar pumping installations were designed backwards. Somebody decided how many panels the budget would carry, picked a pump that looked about right, and then discovered on commissioning that the system delivers plenty of water in June and nowhere near enough in December.

The design sequence that works runs the other way, and it only has one honest starting point: how much water is needed, on the worst day of the year, at the height it has to be lifted. Everything else in the system falls out of that number. The array is the last thing you size, not the first.

This article sets out that sequence end to end. We have covered the equipment itself in our guide to solar water pumping systems and pump selection across the 2 to 200 HP range in the solar pump sizing guide. This is the design method that connects them: demand, head, energy, array, storage.

Step One: Water Demand, and the Design Month

Water demand looks like the easy part. It is where most of the error enters.

For irrigation, demand is driven by crop water requirement, which varies by crop, growth stage, soil and climate, and it peaks in the hottest, driest part of the season. For community water supply, demand is driven by population, service level and livestock, and it is more stable across the year but grows over the design life. For both, the figure that matters is the peak daily demand, not the annual average.

The critical move is choosing the design month. Solar irradiance is lowest in winter, but agricultural water demand is usually highest in summer, which is convenient. Community water supply demand is roughly flat, which means the design month is simply the month with the lowest irradiance. Where demand and irradiance both vary, the design month is the one with the worst ratio of available solar energy to required water, and it is frequently not the month anybody would have guessed.

Designing to an annual average irradiance figure is the single most common error in this field. It produces a system that meets demand on paper and under-delivers for three or four months a year, which is exactly when a farmer or a clinic will judge it.

Add a realistic allowance for growth and for losses in distribution. A system sized precisely to today's demand with no margin is a system that fails the moment a household connects or a field is added.

Step Two: Total Dynamic Head

Total dynamic head is the real work the pump must do, and it is considerably more than the depth to water.

Static lift is the vertical distance from the pumping water level to the point of discharge. Note that this is the pumping water level, not the static water level measured in a rested borehole.

Drawdown is the drop in water level inside the well once pumping starts, and it is a property of the aquifer and the well, not of the pump. It has to come from a pumping test. Assuming drawdown, or ignoring it because the static level looked comfortable, is how systems end up running dry at midday.

Friction losses through the rising main, pipework, valves and fittings depend on flow rate and pipe diameter, and they rise with the square of velocity. This is where a design saves or wastes real money: increasing the rising main by one pipe size often cuts friction loss enough to reduce the array size, and the extra pipe is cheaper than the extra panels.

Discharge pressure is whatever the system needs at the outlet, which for a tank inlet is close to nothing and for a drip irrigation network or a pressurised distribution system can be a substantial addition.

Add these together and the result is total dynamic head in metres. A borehole with 40 m to the pumping water level, 8 m of drawdown, 25 m of elevation to the tank, 6 m of friction and 2 m of discharge requirement is an 81 m system, not a 40 m one, and the difference is the whole design.

The design sequence in the only order that works. Systems that start from the panel budget under-deliver in the design month. Source: MIMAH engineering sizing method.

Step Three: Hydraulic Energy

With demand and head established, the hydraulic energy required per day follows directly. Lifting water is one of the few engineering problems where the physics is genuinely simple: energy is proportional to volume multiplied by head.

The daily hydraulic energy in watt-hours is the volume in cubic metres, multiplied by the total dynamic head in metres, multiplied by the density of water and gravitational acceleration, divided by 3,600. In practice engineers use the shortcut that 1 cubic metre lifted 1 metre requires approximately 2.725 watt-hours of hydraulic energy.

So a system delivering 60 cubic metres a day against 81 m of head needs roughly 13,200 watt-hours of hydraulic energy per day. That is the water's requirement. It is not yet the array's requirement, because nothing in the chain is free.

Step Four: System Efficiency

Between the panel and the water there are several losses, and they multiply rather than add.

The pump and motor set converts electrical energy to hydraulic energy at an efficiency that depends heavily on whether it is operating near its best efficiency point. A well-matched submersible pump and motor combination will commonly sit somewhere in the region of 35 to 55 per cent overall for smaller units and higher for larger ones, but the number to use is the one from the manufacturer's curve at the actual duty point, not a rule of thumb.

The controller or inverter takes its own cut, typically in the region of 90 to 97 per cent depending on type and loading, a subject we cover in detail in our article on solar pump VFDs and controllers.

The array itself loses output to temperature, soiling, mismatch, wiring and ageing. Temperature is the big one in hot climates: module output falls as cell temperature rises above the standard test condition of 25 degrees Celsius, and cell temperatures well above ambient are normal in the field. Soiling in dusty conditions is not trivial either. A combined derate in the region of 20 to 30 per cent from nameplate is realistic in the environments where most of this equipment is installed, and being optimistic here is how a system ends up undersized.

Multiply the hydraulic energy by the reciprocal of the combined efficiency and you have the electrical energy the array must deliver on the design day.

Step Five: Peak Sun Hours and Array Size

Peak sun hours is the conversion between daily solar energy and array rating. It is defined as the number of hours per day for which irradiance would need to be 1,000 watts per square metre to deliver the same total energy as the actual day. A location with 5.5 peak sun hours in the design month receives 5.5 kilowatt-hours per square metre per day on the array plane.

The array rating in watts peak is the required daily electrical energy divided by the peak sun hours for the design month, at the design tilt and orientation. Note the qualifier: peak sun hours depend on how the array is mounted, and a tilt optimised for annual yield is not the tilt that maximises output in the worst month. For a system whose design month is in winter, a steeper tilt than the annual optimum will often reduce the required array size.

Use site-specific irradiance data for the design month rather than a national average. Sudan, Nigeria and Egypt all span a considerable range of irradiance and seasonality, and the difference between a coastal and an inland site in the same country can be material.

The result is the minimum array size. Round up to the nearest sensible module configuration that also satisfies the controller's voltage window, which is a real constraint and not a formality: the string voltage has to stay within the controller's operating range across the full temperature swing the site will see, including cold mornings when open circuit voltage is at its highest.

The full arithmetic chain for a community supply, from population to array rating. Note the head is 77 m, not the 52 m somebody would quote from the water level. Source: MIMAH worked example.

Step Six: Storage, Not Batteries

The instinct to add batteries to a solar pumping system should be resisted in almost every case. Batteries add cost, add a maintenance burden, add a failure mode, and have a service life measured in a handful of years.

Water storage does the same job better. A tank stores the day's yield for use overnight and through cloudy periods, it has no moving parts, it requires almost no maintenance, and it will still be working in twenty years. The design convention is to size storage for one to three days of demand depending on how critical continuity is and how variable the weather is, with community water supply and clinical applications sitting at the higher end.

The tank also has to sit high enough to deliver the required pressure at the point of use by gravity, and that elevation is part of the total dynamic head calculated in step two. This is the circularity that catches out first-time designers: raising the tank to improve distribution pressure increases the head, which increases the array. It is worth checking whether a lower tank plus a small booster arrangement is cheaper overall.

A Worked Example

Consider a community supply for 1,500 people at 30 litres per person per day, plus 200 head of livestock at 25 litres each. Daily demand is 45 cubic metres for people and 5 for livestock, so 50 cubic metres, and adding 10 per cent for distribution losses and growth gives a design figure of 55 cubic metres a day.

The borehole has a pumping water level of 52 m under the intended rate, the tank inlet sits 18 m above the wellhead, friction through the rising main and delivery line is 7 m at the design flow, and the tank inlet requires no additional pressure. Total dynamic head is 77 m.

Hydraulic energy is 55 multiplied by 77 multiplied by 2.725, which is approximately 11,540 watt-hours per day.

Assume a pump and motor efficiency of 45 per cent at the duty point and a controller efficiency of 95 per cent. Electrical energy required at the array output is 11,540 divided by 0.4275, or approximately 27,000 watt-hours per day.

With a design month peak sun hours figure of 5.2 and a combined array derate of 25 per cent, the array rating is 27,000 divided by 5.2, divided by 0.75, which is approximately 6,920 watts peak. In practice that becomes a 7 kWp array built from whatever module the supply chain can support consistently, configured to keep string voltage inside the controller window.

Storage at two days of demand is 110 cubic metres, which in most schemes means more than one tank and a decision about whether they sit in parallel at the same level or in a staged arrangement.

Note what happened in that example: the head was 77 m, not the 52 m somebody would have quoted from the water level, and the array requirement scaled with it.

Where Designs Go Wrong

Using static water level instead of pumping water level. This understates head, undersizes the array, and produces a system that cannot deliver its rated flow. It is the most frequent single error in the field.

Designing to average irradiance. Produces seasonal failure, and the season it fails in is the one the client remembers.

Optimistic efficiency assumptions. Each optimistic figure is individually defensible and they compound. Three cheerful assumptions of 10 per cent each remove roughly a quarter of the system's capacity.

Ignoring the controller voltage window. A correctly sized array in the wrong string configuration will not run, and will sometimes only reveal the problem on the coldest morning of the year.

Sizing the rising main by habit. Pipe is the cheapest place to buy back head. Checking one size up is a five minute calculation that regularly pays for itself several times over in reduced array cost.

Skipping the pumping test. Drawdown is not a value that can be estimated responsibly, and a system designed without it is a guess with an invoice attached.

Why the design month, not the annual average, sets the array. This single error accounts for most seasonally failing pumping schemes. Source: MIMAH engineering sizing method.

Frequently Asked Questions

What is total dynamic head in a solar pumping system? It is the sum of static lift from the pumping water level to the discharge point, well drawdown at the design flow, friction losses through pipework and fittings, and any pressure required at the outlet. It is the head the pump actually works against, and it is always larger than the depth to water.

How many peak sun hours should I design for? Use the figure for the design month at your site and your array tilt, not the annual average and not a national figure. The design month is the one with the worst ratio of available solar energy to required water.

Do I need batteries in a solar pumping system? Usually not. Water storage in a tank achieves the same continuity at lower cost, with far less maintenance and a much longer service life. Batteries are justified mainly where the pumped water cannot be stored or where the same array serves other loads.

Is there a standard for solar pumping systems? Yes. IEC 62253 covers design qualification and performance measurement for photovoltaic pumping systems, and it is the reference point for testing claims about equipment performance.

The Design Is Only As Good As the Site Data

Every step in this sequence depends on numbers from the site: the pumping test, the irradiance in the design month, the real demand rather than the assumed one, and the actual pump curve rather than a category average. A design method applied to guessed inputs produces a confident answer that happens to be wrong.

That is why the survey matters more than the spreadsheet. Across solar pumping and water infrastructure work in Sudan, Nigeria and Egypt, the systems that underperform are rarely the ones where the arithmetic was mistaken. They are the ones where the drawdown was assumed, the design month was skipped, or the demand figure came from somebody's estimate rather than a count.

Planning a solar pumping scheme? Talk to our engineering team. We will work from your pumping test and site irradiance data and give you a design with the assumptions written down, so you can see exactly what the system will and will not do in the worst month.