Technical Guides

Two Fans in Series or in Parallel: What Changes

SXDOOL EC axial cooling fan — series and parallel fan arrangements

Ask for more cooling and you will be offered a larger fan, a faster fan, or another fan. The third option is the one that most often disappoints, and the reason is rarely the fan itself. It is the arrangement. Two identical units can be mounted so that they reinforce each other, or so that one of them does most of the work while the other idles, and from the outside both installations look equally sensible.

The vocabulary borrowed from electrical circuits — series and parallel — is accurate about the topology and unhelpful about the consequence. In a circuit the two arrangements change voltage and current in ways that are easy to picture. In an air path the same two topologies change pressure and flow, and the result depends on something the topology alone does not describe: the impedance of whatever the fans are attached to.

1. What series actually does

Series means the discharge of one fan feeds the intake of the next, so the air passes through both in order. Each unit adds the pressure it can generate while passing the flow it is given. The pressures add. The flow does not multiply. The pair moves roughly what a single unit would move through that same path, but it can push that flow against a more resistant path than a single unit could manage.

Series is therefore a remedy for pressure, not for volume. If the complaint is that air cannot be forced through a dense fin stack, a long duct, or a filter that loads up over time, series is the correct instinct. If the complaint is that the enclosure is simply too warm and the path is open, series is the wrong tool applied confidently, which is worse than being wrong hesitantly.

There is a second condition that series imposes and that drawings usually omit. The downstream unit must be fed with air that is still coherent. Two impellers mounted flange to flange with nothing between them put the second impeller directly in the wake of the first, in air that is spinning and turbulent rather than axial. The pair then delivers less than the sum its curves promise, and the shortfall grows as the units are brought closer together.

2. What parallel actually does

Parallel means both units discharge into a common volume, side by side. Their flows add at whatever pressure both can develop. The pair moves more air than one unit, but only at a pressure the weaker of the two can sustain, and only if the path is open enough to accept the extra volume.

Parallel is a remedy for flow at low resistance. It suits an enclosure with generous openings, a rack front with a clear intake field, or a heat sink with wide fin spacing. It does not suit a path that is already restrictive, because a restrictive path converts the additional flow into additional pressure that the fans cannot use, and the extra unit spends its energy heating air that never travels anywhere useful.

3. The system curve decides, not the topology

Neither description means anything until the fan curve and the system impedance are drawn on the same axes and their intersection is read. The fan curve alone describes a capability; the impedance curve describes a demand; the intersection is the only point at which the two agree. Both curves are worth understanding before an arrangement is chosen: see reading airflow against pressure, calculating system impedance, and reading a pressure–flow diagram properly.

Seen this way, series moves the operating curve upward — more pressure available at any given flow. Parallel extends it rightward — more flow available at any given pressure. Whether either movement reaches the operating point that matters depends entirely on the slope of the system curve. Against a steep curve, the upward move is the one that pays. Against a flat curve, the rightward move is. Against a curve that is steep in one region and flat in another, the answer changes with the season, the filter loading, and the fan speed, which is why a single arrangement chosen once at the design stage can be right in January and wrong in July.

4. The series failure that looks like success

Series fails quietly in open paths. The second unit adds pressure the installation never asked for, the pair draws roughly twice the power, and the measurable increase in through-flow is small enough to be lost in the noise of a temperature reading. The installation looks correct, the current draw is healthy, the fans sound busy, and the enclosure stays warm.

It also fails in a second, less obvious way: the downstream unit can be pushed toward the free-delivery end of its own curve, where its motor carries the highest load. The pair then runs hotter than either unit would alone, and the failure appears months later as a bearing noise or a seized motor rather than as an immediate thermal problem.

5. The parallel failure that looks like success

Parallel fails when the units are not matched. If one unit can develop more pressure than its neighbour, the stronger one raises the pressure in the shared discharge volume above what the weaker one can hold back, and air moves backwards through the weaker unit. Both impellers spin. One of them is now a leak that the other is pressurising. Total through-flow falls below what a single unit would have delivered, and the pair is blamed for being cheap.

This is the same mechanism that makes multi-fan arrays behave worse than their arithmetic, described from the array side rather than the pair side in recirculation and hot-spot bypass in fan arrays. The fix is not a stronger fan. It is matching, separation, or a barrier that stops the shared volume from becoming a short circuit.

6. Geometry decides more than the datasheet does

Series needs a settled approach. A spacer, a short straight section, or a turning vane between the units restores most of what back-to-back mounting throws away, and it is usually cheaper than the second fan. Parallel needs separation at the intake, so that the two units are not competing for the same volume of available air; two impellers close together in a restricted intake field pull against each other and deliver less than their curves suggest. What inlet clearance and guard geometry do to a single fan is set out in installation effects on real airflow, and the same penalties apply twice over in a pair.

7. Matching, and what “identical” buys you

In series, the weaker unit limits the pair: pressure is added, but the flow both units see is set by the unit that saturates first, and a mismatched pair can put the stronger unit into a stalled region it was never selected for. In parallel, the weaker unit becomes the return path described above. Identical model, identical speed setting, identical voltage tap, and identical age is the only configuration in which the simple arithmetic holds.

Where speed control is available, independent control per unit is worth more than a shared command, because a single command delivered to two units that are not identical reproduces exactly the pressure imbalance the pairing was meant to remove.

8. Redundancy is a different question

Two fans in parallel are not automatically a redundant pair, and two fans in series are not automatically a reliable one. Redundancy asks what happens after a unit stops: in a parallel pair a failed unit becomes an open return path unless something closes it, and in a series pair a failed unit becomes a blockage that the survivor cannot push through. Both are solvable, but they are solved with flaps, blanking, or sensing — not with the arrangement itself. The reliability layer belongs on top of the airflow layer, and it is treated separately in redundant fan configurations for mission-critical racks.

9. Three questions that pick the arrangement

First: is the path restrictive or open? A loaded filter, a dense fin stack, a long duct, or a narrow plenum entry all mean the installation is pressure-limited, and series is the arrangement that speaks to pressure. A clear intake field and generous discharge mean the installation is flow-limited, and parallel is the arrangement that speaks to flow.

Second: does the temperature respond to flow at all? If raising the speed of the existing fan does not move the reading, more flow in any arrangement will not move it either, and the problem is the path rather than the capacity. That test costs nothing and settles most arguments before a purchase order exists.

Third: is the problem global or local? A uniformly warm enclosure is a capacity question. A single hot component in an otherwise acceptable enclosure is a path question, and neither series nor parallel answers it — a baffle does. Arrays that are already dense enough to have hot spots inside them are covered in airflow density in integrated fan arrays.

10. What this changes when you specify

Specify the path, not the count. A request that names the number of fans has already chosen the arrangement before anyone has looked at the impedance, and the arrangement is the part that decides whether the second unit helps. A request that describes the intake, the discharge, the obstruction between them, and the thermal target leaves the arrangement open, which is where it belongs until the curves have been read.

Ask what the second unit is for. If the answer is pressure, series with a settled approach. If the answer is volume, parallel with separated intakes and matched units. If the answer is “because the first one was not enough”, the honest reply is that the first one was probably asked to solve a path problem, and a second one will be asked to solve it twice.

And treat the pair as one component with one operating point, not as two components with two datasheets. The curves add in a specific way, the power adds in a way that is rarely proportional to the benefit, and the noise adds in a way that neighbours notice before the thermocouple does. That is the whole argument for reading the arrangement before ordering it.

If you are reworking an enclosure that already has a pair of fans in it, send us the arrangement, the spacing, and the obstruction between the units along with the thermal target. The RFQ guide lists what we need in order to answer the arrangement question rather than only the airflow question.