Pump Sizing in Heating Networks: Pump Head, Pump Power, Pump Energy, Validation

Network pumps keep the water in a heating network moving and must provide sufficient differential pressure at every building. Directly from the hydraulically calculated network design, nPro determines the critical point, the required pump head, the design flow rate, the electric pump power and the pump electricity demand as an hourly profile over the entire year. This applies to the central pumping concept of conventional local and district heating networks as well as to passive 5GDHC networks, in which decentralized pumps in the buildings move the water. On this page you will find the calculation principles, the assumptions and a comparison with the design tool Grundfos Product Center.

Pumping concepts: central network pump or passive 5GDHC network

nPro distinguishes between two pumping concepts, which are selected in the project:

Central pumping concept: A network pump in the energy hub circulates the water through the entire network. A differential pressure is available at the substations and is reduced by control valves. This is the standard in local district heating and district heating networks.
Decentralized pumping concept (passive 5GDHC network): The network has no central network pump. Instead, each substation moves the water with its own pump in its respective flow direction: from supply to return when drawing heat, and from return to supply when feeding in. This concept is referred to as a passive 5GDHC network. nPro calculates 5GDHC networks with a central network pump according to the central pumping concept.

Central pumping concept: network pump in the energy hub

Critical point

The critical point is the substation with the lowest available differential pressure between supply and return. nPro determines it from the node pressures of the hydraulic network calculation, not from the longest pipe route. This means the critical point is identified correctly in meshed networks and with unevenly distributed loads as well, where it is not necessarily located at the most distant building. Pure feed-in units without their own heat demand cannot be the critical point.

Pump head of the network pump

The pump head of the network pump consists of four components:

\Delta p_P = (1 + z_{fit}) \cdot \Delta p_{pipe} + \Delta p_{sub} + \Delta p_{hub}

with $\Delta p_{pipe}$ as the pressure loss of the pipes in supply and return between the energy hub and the critical point, $z_{fit}$ as the surcharge for fittings and valves (project input in % of the pipe pressure loss), $\Delta p_{sub}$ as the minimum differential pressure of the substation at the critical point (project input) and $\Delta p_{hub}$ as the pressure loss of the energy hub.

Geodetic height differences are not included in the pump head, as they cancel out between supply and return in the closed circuit. However, they determine the pressure level in the network and are therefore relevant for pressure maintenance and the pressure rating of the pipes.

Design flow rate and design case

The network pump is sized for the design flow rate at the energy hub, which is also the basis of the pipe sizing. Only the flow rate delivered by the energy hub into the network is taken into account. If, at times of feed-in surplus, water flows in the opposite direction towards the energy hub, this flow is driven by the feed-in pumps (see below).

For the pressure loss of the pipes, nPro compares two states: the static design case of the pipe sizing and the time step with the highest load from the dynamic annual simulation. The larger pressure loss governs. In two cases, only the simulation state is used:

Relevant reverse flow at the energy hub: If the feed-in surplus at the energy hub exceeds 25 % of its maximum heat delivery, the static design case is no longer representative. The same applies if no detailed pipe sizing has been carried out.
Meshed networks: If the node pressures of the design case are not consistent in a meshed network, the pressure loss from the simulation is scaled up to the design flow rate and a warning is issued:

\Delta p_{pipe} = \Delta p_{pipe,sim} \cdot \left( \frac{\dot{V}_{design}}{\dot{V}_{sim}} \right)^{1.8}

The exponent of 1.8 lies between hydraulically smooth pipes (1.75 according to Blasius) and fully rough flow (2.0) and represents the transition range in which district heating pipes are usually operated.

Pump power and efficiency

The hydraulic and electric power of the pump at the design point are:

P_{hydr} = \Delta p_P \cdot \dot{V}_{design} \qquad P_{el} = \frac{P_{hydr}}{\eta}

The efficiency $\eta$ is the overall efficiency of hydraulics, motor and electronics. For large district heating pumps, it is typically between 65 % and 85 %; nPro uses 75 % as the maximum value. Small pumps reach considerably lower efficiencies, which is why the efficiency is calculated depending on the hydraulic power:

\eta = \max \left( \eta_{max} \cdot \frac{P_{hydr}}{P_{hydr} + P_{1/2}} ,\; \eta_{min} \right)

with $\eta_{max} = 0.75$ , $P_{1/2} = 0.03$ kW as the hydraulic power at which half of the maximum value is reached, and $\eta_{min} = 0.2$ as the lower limit. The same curve applies to network pumps, feed-in pumps and the building pumps in passive 5GDHC networks.

Table 1: Overall pump efficiency in nPro depending on the hydraulic power.

Hydraulic power	10 W	30 W	100 W	1 kW	10 kW
Overall efficiency	20 %	38 %	58 %	73 %	75 %

Feed-in pumps

Secondary energy hubs and prosumers that feed waste heat from the return into the supply require their own feed-in pumps. Their pump head results from the pressure difference between return and supply at their network node plus their internal pressure loss (energy hub or substation):

\Delta p_{feed} = \max \left( p_{ret} - p_{sup},\; 0 \right) + \Delta p_{int}

In addition, a summer time step with feed-in surplus is considered, as the feed-in pumps then work against the most unfavourable pressure conditions. The network pump only circulates the share that the secondary energy hubs do not cover themselves.

If, on the other hand, prosumers feed in from supply to return, they draw supply water like a consumer and return it heated into the return pipe. No feed-in pump is needed in this case; the network pump also circulates the feed-in flow.

Decentralized pumping concept: passive 5GDHC network

In passive 5GDHC networks there is no central network pump. The energy hub is passive; its pressure loss is overcome by the pumps in the substations. Each substation receives one pump per flow direction, so prosumers with both heat draw and feed-in may have two.

The pump head of a station results from the pressure difference between supply and return at its node that it has to overcome in its flow direction. If this is negative, there is a pressure surplus at the station that is throttled, and the station does not need a pump. For this, nPro evaluates up to three network states: the static design case, the time step with the highest load in winter, and a summer time step with feed-in surplus. A station is sized only for states in which it is actually in operation; the largest required pump head governs.

The design flow rate of each station results from its design load and the temperature difference in its peak hour:

\dot{V}_{design} = \frac{\dot{Q}_{design}}{\rho \cdot c_p \cdot \Delta T}

For heat draw, the design load is the larger of the maximum hourly load and the static heat load of the building, without a diversity factor, since each building has its own pump. Density and heat capacity are based on water or the selected water-glycol mixture.

The calculated pump head covers the network side only. Pressure losses inside the building, for example in the heat exchanger or the evaporator of the heat pump, are not included and must be added when selecting the pump. If the pump head of a station exceeds the maximum pump head of the building pumps set in the project, nPro issues a warning for that building.

Secondary energy hubs such as borehole fields are considered passive if their remaining pump head is no more than 5 kPa (0.05 bar). If they require more, nPro issues a warning that they could not fully feed in their share without their own pump. In passive 5GDHC networks, the building with the largest pump head of a heat draw pump is shown as the critical point. More on the design of 5GDHC networks can be found under sizing 5GDHC networks.

Pump electricity demand

The pump electricity demand is calculated hourly over the entire year. The flow rate of each hour follows from the load and the temperature difference in that hour; the pump head is assumed to be constant (speed-controlled pump with constant pressure control):

\dot{V}(t) = \frac{\dot{Q}(t)}{\rho \cdot c_p \cdot \Delta T(t)} \qquad W_{el} = \frac{\Delta p_P \cdot V_a}{\eta}

with $V_a$ as the annually circulated water volume. Pumps with critical-point control reduce the pump head at part load, so the calculated electricity demand tends to be on the safe side. The hourly profile of all pumps is passed to the energy hub simulation as an electricity demand. In addition, nPro reports the pump electricity demand relative to the delivered heat and cold in %.

Results of the pump sizing

Table 2: Results of the pump sizing in nPro.

Result	Description
Critical point	Building with the lowest available differential pressure (central network pump) or the largest pump head (passive 5GDHC network)
Pressure loss	Split into pipes, fittings, substation and energy hub
Mean pressure gradient	Pipe pressure loss per metre of pipe length to the critical point in Pa/m, branched networks only
Design flow rate	Flow rate of the network pump or sum of all pumping stations in m³/h
Pump power	Hydraulic and electric power at the design point in kW, for the network pump with a central network pump, as the sum of all building pumps in passive 5GDHC networks
Pump electricity demand	Annual electricity demand of all pumps in MWh/a, relative to the delivered heat and cold in %, and as an hourly profile
Values per building	Pump head, flow rate, electric power and electricity demand of the building or feed-in pump, as well as the available differential pressure

Assumptions and limitations

Geodetic height differences are not included in the pump head, as they cancel out in the closed circuit.
The pump head is constant over the year; the flow rate follows the load hour by hour.
Transient effects such as pressure surges when starting pumps or closing valves are not considered.
nPro outputs the design point consisting of pump head and flow rate. The selection of a specific pump model based on its pump curve takes place in the subsequent planning.

Validation of the pump sizing

For validation, the network pump of a real neighbourhood network in Rhineland-Palatinate was sized with the design tool Grundfos Product Center (as of September 2026) and compared with the results of nPro. Both calculations are based on the same building loads and nominal pipe diameters. Boundary conditions: meshed network with 14 sections and 11 supply zones, sum of peak loads 4,421 kW including one zone with 1,169 kW heat feed-in, supply temperature approx. 85 °C, return temperature approx. 42 °C. In Grundfos, a pipe roughness of 0.25 mm, a differential pressure of 0.4 bar at most substations and no additional local losses were applied. The pump there is controlled on the differential pressure in the supply zones.

Table 3: Comparison of the pump sizing between nPro and Grundfos Product Center (as of September 2026) for a meshed neighbourhood network with 11 supply zones. Pipe pressure loss for Grundfos derived from the pump head minus 0.4 bar differential pressure at the critical point; conversion from m to bar with 983.2 kg/m³ (60 °C).

Parameter	nPro	Grundfos Product Center	Deviation
Flow rate of network pump	90 m³/h	89.98 m³/h	0.0 %
Pipe pressure loss	1.16 bar	1.29 bar	−10 %
Total pump head	1.95 bar	1.69 bar (17.52 m)	+15 %
Overall efficiency at design point	74 %	69.6 %	+4.4 pp
Electric power at design point	6.6 kW	6.06 kW	+9 %
Annual pump electricity demand	15.04 MWh	15.40 MWh	−2.3 %

The design flow rate matches almost exactly. The 15 % higher pump head in nPro is mainly due to the surcharges: nPro additionally takes into account 0.2 bar for the energy hub and 25 % for fittings (0.29 bar), which were not applied in Grundfos. Conversely, 0.3 bar instead of 0.4 bar was set for the substation in nPro. With the same assumptions as in Grundfos, nPro would yield a pump head of 1.16 bar + 0.4 bar = 1.56 bar, just under 8 % less than in Grundfos. The pure pipe pressure loss in nPro is 10 % lower. Differences arise here from, among other things, the pipe roughness, the fluid properties and the distribution of flow rates in the meshes of the network.

At 74 %, the overall efficiency from the nPro efficiency curve is close to the value of the pump selected by Grundfos (69.6 % including motor and frequency converter). Due to the higher pump head, the electric power at the design point is 9 % higher in nPro.

The annual pump electricity demand of the two calculations differs by only 2.3 %. However, this is the result of two opposing effects. Grundfos reduces the pump head at part load, to 47 % of the design pump head at 25 % flow, but calculates with a generic load profile of four load levels in which the pump circulates around 360,600 m³ per year. nPro calculates with a constant pump head, but with the hourly load profile of the buildings, in which around 207,000 m³ are circulated per year. Per cubic metre circulated, nPro is therefore on the safe side, as expected from the constant pump head.

For this operating point, Grundfos selects two speed-controlled TPE3 80-210 pumps with 3 kW each in parallel operation. nPro outputs the operating point consisting of pump head and flow rate; the selection of a specific pump model and the split across several pumps take place in the subsequent planning.

Further comparisons of nPro with established planning tools can be found under validation.

Sources

EnergieSchweiz / Verenum: Planungshandbuch Fernwärme (district heating planning handbook), version 1.2.
Grundfos: Grundfos Product Center, pump sizing software. Version 2026.39.000, as of September 2026.