Solar Pump at High Altitude: Altitude Does Not Give You More Sun

The short answer: altitude does not give you more sun, and it does not make a submersible pump weaker. Altitude removes atmospheric pressure, and atmospheric pressure is the only thing that lets a pump lift water by suction. At sea level the theoretical suction limit is 10.3 m and the practical limit is about 6 m. At 4,000 m those become 6.3 m and about 2 m. A submersible pump sitting in the water is not affected at all, which is why high ground is submersible country. The second surprise is that “high altitude” and “lots of sun” are not the same thing: in the ten Andean sites I pulled from NASA POWER, the highest station is not the sunniest one, and the sunniest month can deliver 40 percent more than the worst.

I get inquiries from Peru, Bolivia, Chile and northern Argentina every week, and the assumption behind almost all of them is the same: at 4,000 m the sun is stronger, so the pump will be easier. The first half is not always true and the second half is only true if you choose the right kind of pump. This article deals with both, because they are different problems that get confused with each other.

What altitude really changes (and what it does not)

There are exactly three physical quantities that change with altitude and matter to a pumping system.

And here is what does not change, which is just as important:

• The pump curve. A pump curve is head against flow. Head is measured in metres of water, flow in cubic metres per hour, and neither depends on the air pressure above the well. A 4DSC6-101-110-1100 that delivers a given flow at 60 m at sea level delivers the same flow at 60 m at 4,000 m.
• Water density. Fresh water is 1,000 kg/m3 at sea level and at 4,000 m. The work of lifting 1 m3 by 10 m is identical.
• Solar panel rating. A panel rated 550 W is rated at 1,000 W/m2 irradiance and 25 C cell temperature, which is a lab condition, not an altitude condition. What altitude changes is the irradiance you actually receive and the cell temperature you actually run at, and both of those are second-order effects compared with cloud.

The practical consequence is a clean design rule that saves people a lot of money: on high ground, never lift water by suction. Put the pump in the water and the altitude problem disappears entirely. Everything else on this page is about choosing the array and protecting the surface equipment.

The suction lift you lose above 2,000 m

A centrifugal pump that sits above the water surface does not pull water up. Nothing pulls. The atmosphere pushes down on the surface of the well, and the pump lowers the pressure inside the suction pipe so the atmosphere can push water into it. The maximum height that push can reach is fixed by physics:

Maximum suction lift (m) = atmospheric pressure head (m) minus vapour pressure head (m)

At sea level, atmospheric pressure of 101.3 kPa equals 10.3 m of water. At 4,000 m, 62 kPa equals 6.3 m. That is the absolute ceiling, with a perfect pump, zero friction and water at 20 C. Then you subtract the real-world losses. Using a small centrifugal or self-priming pump with an NPSH requirement of 2.5 m, 1.0 m of friction and fittings in the suction line, 0.24 m of water vapour pressure and a 0.5 m safety margin, the numbers look like this:

(Pressure values from the standard atmosphere and from IEC 60664-1 Table A.2, which gives 80.0 kPa at 2,000 m, 70.0 kPa at 3,000 m and 62.0 kPa at 4,000 m. Practical column assumes NPSH requirement 2.5 m, suction line loss 1.0 m, safety margin 0.5 m, water at 20 C.)

Read the last column again. A pump that a catalogue calls a 7 m self-priming pump is a 2 m pump at 4,000 m. That is not a small correction, it is a factor of three, and it is the single most common reason a surface pump that worked perfectly in a valley “does not work” when it is installed on the plateau.

Two things make it worse in practice:

1. The catalogue number is a sea-level number. Pump catalogues are written for a reference atmosphere. When a model is listed with a suction lift of 7 m or 8 m, that is measured at roughly 100 kPa. Nobody restates it for your elevation.
2. Suction leaks matter more. Any tiny air leak in a suction flange, a worn foot valve or a joint that would be cosmetic at sea level becomes a performance failure at 3,500 m, because there is less pressure to push water past it and cavitation starts sooner.

The fix is not a bigger surface pump. A bigger pump does not raise the pressure of the atmosphere. The fix is to shorten the suction lift to a metre or two, or to put the pump in the water.

Why a submersible ignores altitude and a surface pump does not

This is the design decision that separates a high-altitude system that works from one that keeps tripping on dry-run protection.

A surface pump does two jobs: it has to lift water from the source up to its own impeller, and then push it to where you need it. Only the first job depends on air pressure. A submersible does only the second job, because it is already sitting in the water. Its entire rating, every metre of head and every cubic metre per hour, is about pushing.

Our own catalogue shows the two versions of the same hydraulics side by side. The DQD series is the submersible version and the HJQD series is the surface centrifugal version, and they share the same nine hydraulic ratings:

Same head, same flow, same panel count, and one of them is unusable the moment the water level drops 3 m below the bank. On the altiplano, where dry-season water levels move and hand-dug wells sit right next to the pump shed, that is the difference between a working system and a returned one.

The one case where a surface pump still makes sense at altitude is when the water is genuinely at your feet: a canal, a spring box, a storage tank, an irrigation ditch. Then the suction lift is under a metre and the altitude penalty is irrelevant. That is exactly the case for the DCPM15-14-48-550 (550 W, 48 V, 15 m3/h at 14 m) or the DQB3.0-65-72-750 (750 W, 72 V, 3 m3/h at 65 m) sitting a few metres from a canal, and for the HJET2.7-45-48-550 (550 W, 48 V, 2.7 m3/h at 45 m) on a spring box. What you should not do is park a self-priming pump on the bank of a 3,900 m well and expect it to lift 6 m.

Higher is often cloudier: the data that breaks the rule of thumb

Now the part that costs money. Everyone from a buyer to a local installer will tell you that high altitude means strong sun. I pulled monthly climatology from NASA POWER for ten Andean locations, and the data does not support it.

(NASA POWER climatology of all-sky surface shortwave downward irradiance, kWh/m2/day, monthly averages.)

Three conclusions follow from that table.

1. Altitude does not predict your solar resource. Cerro de Pasco sits at 4,380 m, the highest site in the list, and averages 4.97 kWh/m2/day. Arequipa sits 2,045 m lower and averages 6.63. That is a 25 percent advantage for the lower site. Meanwhile Uyuni, at 3,670 m, beats El Alto at 4,150 m by 12 percent even though it is 480 m lower. What decides these numbers is cloud: the dry altiplano around Uyuni, Potosi and Puno is one of the clearest places on earth, while the wet puna of central and northern Peru, Cerro de Pasco, Huancayo, Cajamarca, Cusco, is under heavy seasonal cloud. Two plateaus at the same altitude, one 35 percent richer than the other.

2. The minimum month is not the month you expect. On the dry altiplano the worst month is June, the austral winter: the sky is at its clearest but the sun is at its lowest and the days are shortest. Uyuni swings from 8.38 in November down to 5.00 in June, a 40 percent drop. In the wet puna the worst month moves to March, at the end of the rainy season, because cloud is the controlling factor there rather than sun angle. Cerro de Pasco and Cajamarca both bottom out at 4.42, in March.

3. The dry season and the weak-sun season are the same season on the altiplano. June and July are when irrigation demand for forage is highest in the Bolivian and southern Peruvian highlands, and they are also the lowest-output months of the year. An array sized on the November number will be 40 percent short exactly when the farmer needs it. Size on June.

The method is free and it takes five minutes. Look up your own latitude and longitude in NASA POWER climatology, take the worst month, and design the array on that number. If you want the deeper version of this method, it is written out in the Africa guide, where the mechanism is rain rather than altitude but the arithmetic is identical.

The worst month is not the same month everywhere on the plateau

Because hundreds of solar pumping projects in Peru, Bolivia and Chile are sized on an annual average, they all fail in the same few weeks, and the weeks differ depending on which side of the Andes you are standing on.

The coastal column is the extreme case, and it is a reminder that altitude is not the only thing that matters: at Arica the June value of 3.25 is lower than any month at any highland station in the table, because winter brings the camanchaca, the coastal fog that sits over the Atacama for weeks. If you are pumping for an olive or avocado farm on the coastal plain below the Andes, you are in a harder solar climate than a farmer at 4,000 m.

The practical rule for the plateau: take your worst month, multiply your daily volume by 1.3, and size on that. The 1.3 factor covers the panel soiling and the ageing that this dry, dusty, high-UV environment delivers faster than a temperate site. The full sizing arithmetic is in how to size a solar water pump.

Both ends of the day: a cold dawn and a hot noon

High altitude has the widest daily temperature swing of any inhabited place on earth. El Alto swings from around minus 3 C before sunrise to 15 C in the afternoon, and the daily range of 18 to 20 C is normal. That gives you two different electrical problems in the same day, and both are checked against the controller.

The cold dawn is the dangerous end. Panel open-circuit voltage rises as temperature falls, by roughly 0.3 percent per degree below 25 C. Take a 550 W panel with a nameplate VOC of about 49.5 V and a nameplate Vmp of about 42 V, and put it at 3,900 m on a July morning at minus 10 C. That is 35 degrees below 25 C, so nameplate voltage rises about 10.5 percent:

That last row is the one that catches people. The standard advice in the tropics is that a DF-110 controller with a 220 V ceiling easily takes four panels. In the Andean winter it does not. Three panels in series is the ceiling for a 110 V controller at high altitude, and if you need more array power you add a parallel string rather than a longer series string. The full voltage-drop and string arithmetic is in the cable sizing guide.

The hot noon is the milder end at altitude, which is the reverse of the tropics. Cell temperature depends on irradiance, air temperature and cooling, and here the three effects fight each other: the irradiance is high, the air is thin so convective cooling is weaker, but the air temperature is low. In practice a high-altitude panel runs cooler than a tropical one, so you do not get the tropical failure mode where the string voltage sags toward the bottom of the MPPT window at midday. What you do lose is de-rating headroom: thin air removes roughly 40 percent of the air molecules available to carry heat away from the back of the panel, so a still, high-irradiance afternoon at 4,000 m runs hotter than the ambient temperature suggests. Do not count on altitude to give you extra watts. Plan for the standard 1.3 times pump power in panel nameplate, and read MPPT vs PWM if you want the mechanism.

There is a third electrical item that is pure altitude and has nothing to do with the sun: the maximum installation altitude of your controller. IEC 60664-1, the standard that sets insulation clearances for low-voltage equipment, applies to equipment used up to 2,000 m, and above that it requires the clearance inside the equipment to be increased: a factor of 1.14 at 3,000 m, 1.29 at 4,000 m and 1.48 at 5,000 m. Many controllers, drives and inverters are therefore rated for 2,000 m as standard and either de-rate above that or are not certified at all. Ask the question in writing before you buy: what is the maximum installation altitude of this controller? A controller rated to 2,000 m installed at 4,200 m is a warranty claim waiting to happen, and it is a failure mode no amount of correct plumbing will fix.

Freeze, retreating glaciers and a falling water table

Two slow problems that decide whether a high-altitude system is still running in five years.

Freezing. The submersible itself is safe, because it sits below the water line where the water stays at 4 to 10 C all year. What freezes is everything above the water line:

• The top metre of water in a hand-dug well or a shallow pozo. This is the classic altiplano failure. If the pump sits in the top of the water column, in June and July the impeller can be sitting in ice. Set the pump at least 1.5 m below the lowest expected water level, and more if the well is shallow.
• The drop pipe above the water line, the surface pipe from the well head to the tank, the pressure switch, the hose fittings and the check valve.
• The tank, if it is steel or thin-walled plastic and left full at a head that can freeze solid.

The measures that work are the boring ones: bury the surface pipe below 0.8 m, or better, drain it back after each pumping cycle with a small weep hole so the line empties itself; set the pump deeper; insulate or bury the tank; and keep the controller indoors. This is the same problem set as a temperate market, and it is covered in more detail in solar pumps in winter, but the high-altitude version is harder because the freeze is followed by a 20 C afternoon that thaws the line just in time for it to freeze again the next night.

A falling water table. This one is specific to the Andes and it is not reversible. Around 70 percent of the world tropical glaciers are in the Peruvian, Bolivian and Ecuadorian high Andes, and they are retreating fast. Bolivia Chacaltaya glacier went from 97 percent mass loss between 1960 and 2003 to complete disappearance by 2009. Peruvian glaciers have lost more than one fifth of their mass in roughly 35 years, cutting water flow to the coastal region by about 12 percent. La Paz draws around 30 percent of its water supply from a glacial basin, Quito around 50 percent. The Yale Environment 360 reporting on the World Bank assessment puts the share of Andes tropical glaciers expected to disappear within 20 years at most of them.

For a pump buyer the consequence is concrete: the static water level you measure today is not the static water level your pump will see in ten years. Design for it:

1. Buy head margin, not just adequacy. If your total head is 43 m, choose a pump with a maximum head of at least 65 to 70 m rather than one rated exactly at 45 m. On these pumps, a working point at 90 percent of maximum head is a working point with almost no flow, and it leaves you nothing when the level drops. Anything in the high-head range is fine, this is only about leaving room.
2. Run your riser pipe generously so you can lower the pump. Leave enough cable and enough pipe that the pump can be dropped another 10 or 15 m in a dry decade without a new installation.
3. Measure seasonal, not snapshot. Record the water level in the wet season and in the dry season before you size. In the altiplano the dry-season measurement is the only one that matters.

Worked example: a 3,900 m alpaca and forage farm

A real shape of project from the Bolivian altiplano. A family with 150 alpacas and a 0.1 hectare forage plot, a 4 inch borehole 70 m deep, and a house 60 m from the well head.

Step 1 - the site data, measured in the dry season, not in the wet season.

Step 2 - the daily volume, split into the base load and the irrigation load.

• Household: 0.5 m3/day
• 150 alpacas at about 5 litres per head per day: 0.75 m3/day
• Dry-season forage irrigation: 0.1 ha receiving 60 mm per week is 60 m3 per week. Pump it into a reservoir gradually rather than in one burst, so it costs 60 / 7 = 8.6 m3/day
• Peak daily total: about 10 m3/day

Step 3 - convert to a flow rate using the worst month. June at 3,900 m gives about 5.3 kWh/m2/day, and at 3,900 m the cool air helps the array, so use a system factor of 0.85 rather than the 0.78 you would use in a hot climate:

5.3 x 0.85 = 4.5 effective pumping hours

10 m3/day / 4.5 h = 2.2 m3/h required at 43 m of head

Step 4 - pick the model with margin, not the model that just fits. Do not choose a pump whose maximum head is 45 m for a 43 m job; on a solar pump that is a working point with almost no flow, and it has nothing left for the dry decade. Choose a maximum head of at least 1.5 times your working head, so 65 m or more here.

The 4DSC6-101-110-1100 is the natural fit: 1,100 W, 110 V, 6 m3/h maximum flow and 101 m maximum head. At 43 m of head it delivers comfortably more than the 2.2 m3/h you need, and it has 58 m of head still in reserve for the day the water level drops. Stainless impeller, which is the right call for Andean groundwater that often carries fine glacial silt.

Step 5 - the array and the string. 1.3 x 1,100 W = 1,430 W, so three 550 W panels at 1,650 W. Three in series gives 148 V nameplate VOC and about 164 V on a minus 10 C morning, inside the 220 V limit of the 110 V controller. Nameplate Vmp of the string is about 126 V, inside the controller MPPT window. Four panels in series would reach 219 V and is not acceptable. If you need a fourth panel, add it as a second string in parallel.

Step 6 - storage. A 5 m3 household tank at the house covers two days of domestic and livestock demand, and a separate 60 m3 lined reservoir takes the weekly irrigation slug so the pump never has to run harder than 2.2 m3/h. This is the one place where high altitude is genuinely easier than the tropics: there is no wet-season cloud problem to design around, only the June sun angle, and the water you store never evaporates as fast as it does in the Sahel.

A real model ladder for high-altitude sites

Every model below is from our current catalogue, and every one of them is a genuine product with a published rating. The bracket in the middle is the one to read: it is the maximum head, and your working head should never be more than about two thirds of it.

The pattern of the range is consistent: 2DPC, 3DPC and 4DPC for small bores and clean water, 3DSC and 4DSC where the water carries silt or you need more head from a 4 inch hole, 3DSS screw pumps for the awkward combination of a very small daily volume and a very deep water level, DQD for anything where the water is open and shallow, and the 4/6DSC family with the A/D suffix once the customer is a village rather than a family. The A/D models matter more on the plateau than in most markets, because many highland communities have an intermittent grid connection that can be used in the evening while the panels do the daytime work, without buying a battery. The economics of that choice are in solar pump without battery.

Generalisations worth keeping: a low-head impeller is fine for a DQD in a reservoir and wrong for a 100 m borehole, and if your water is glacial silt rather than clean sand, move up to a stainless impeller and stay there.

Five mistakes that end high-altitude projects

In the order I actually see them:

1. Installing a surface or self-priming pump above the water. This is the number one failure and it is entirely avoidable. A 7 m catalogue suction lift is a 2 m reality at 4,000 m. Put the pump in the water, or move it to within a metre of the surface.
2. Sizing the array on the annual average, or on the plateau average. The worst month runs from 4.42 to 5.54 depending on which side of the Andes you are on, and the swing within one site reaches 40 percent. Design on your own worst month, at your own coordinates.
3. Assuming high altitude means strong sun. Cerro de Pasco at 4,380 m gets 25 percent less annual sun than Arequipa at 2,335 m. Two locations at the same elevation can differ by a third. Cloud decides, not height.
4. Counting the series string at room temperature. A minus 10 C dawn adds about 10 percent to your string VOC. Four 550 W panels on a 220 V controller is 219 V, which is a failure with one volt of margin. Three is the ceiling on a 110 V controller, two on a 48 V controller.
5. Ignoring the falling water table and the freeze. Andean glaciers have lost more than a fifth of their mass in 35 years and the dry season is lengthening. Buy head margin, leave enough riser pipe to drop the pump another 10 m, bury or drain the surface line, and set the pump at least 1.5 m below the lowest water level so it is not sitting in ice in July.

Everything else, the head arithmetic, the panel count, the voltage window, the cable size, the tank volume, is ordinary engineering, and it is all in this article and the ones linked from it.

Planning a solar pump in Peru, Bolivia, Chile or northern Argentina? Put your static water level, your drawdown, your lift and your pipe run into the sizing tool to see your real total head, or message me on WhatsApp with your elevation, static water level, drawdown, lift, pipe run and daily volume, and I will come back with the model, the panel string and the tank size sized for your own worst month.

Related reading: how to calculate pump head for the head arithmetic, how to read a pump curve for why a 45 m pump is not a 45 m pump, solar pump controller wiring diagram for the string and cable layout, and solar pump for village water supply if your project is a community rather than a single farm.