If you have any further questions regarding our services or see potential for cooperation, please contact us by email or phone number given below:
© heatbeat engineering GmbH & heatbeat nrw GmbH 2017-
in the 71st issue of our heatbeat Research Newsletter we present two studies working on the same quantity: temperature spread. The first, an open-access simulation study in Energy Reports, asks how much return temperature can be gained by rebuilding nothing but the plant room of a multi-residential building. The second, a TU Darmstadt preprint, globally optimizes network operation in meshed topologies with several suppliers, on aggregated models of real Darmstadt networks.
The biggest new feature in August concerns pipe selection in the Design Package. In the map view, any combination of pipe types is now available for individual sections and entire routes, organized by manufacturer, product series, and nominal diameter. The same selection applies to automated sizing, where you specify for each run which pipe types the algorithm should use to determine the optimal nominal diameters. You can also manage product series and pipe costs project-wide in one central location via the central assumptions. In addition, we have prepared typical KMR and PMR data sets as manufacturer-neutral records that include costs directly from the KWW technical catalog, enabling an initial cost estimate without the need for additional research.
To help you map your existing networks, we have also refined the management of connection capacities. Based on feedback from several municipal utilities, installed and contractually agreed-upon connection capacities can now be tracked separately, supplemented by a freely adjustable calculation capacity for simulations that may intentionally deviate from reality—for example, in cases of redensification or capacity reductions following building renovations. At the same time, we have updated the data foundations for all heating networks: new pipe costs from the KWW Technical Catalog, current waste heat data from the Waste Heat Platform, an improved source for Germany-wide address data, and higher-resolution elevation data, which are immediately reflected in the revised elevation profile of the design report.
Another new feature is the ability to switch freely between background maps: In addition to OpenStreetMap and the official BKG map series, aerial images of the individual federal states are now available, allowing you to verify route alignments and development directly in the Digital Twin against the actual conditions on the ground. There are also minor improvements to the entry of consumption values, the color coding of feed-in points, and the categories for connection interest. A separate blog post detailing the August update will follow. If you’d like to see the new features applied to your own grid, please feel free to contact us.
High return temperatures raise thermal losses, drive pump electricity and eat into transmission capacity. Above all they limit what can be integrated: large heat pumps, low-temperature waste heat and condensing operation all depend on the sink temperature. How far can the return be lowered without involving residents? The case is an Austrian low-energy building, 64 apartments and 4,840 m² heated area, radiators designed for 60/40 °C, monitored for one year: 74 kWh/m²a in total, of which 45 kWh/m² space heating and 27 kWh/m² hot water including circulation losses. The measured energy-weighted return temperature was 53.5 °C, with 32 % of the energy transmitted above 55 °C. A Polysun model is calibrated against these data; the primary supply is set constant at 90 °C and the load profiles stay unchanged across all variants.
Control optimization alone already works. Relocating sensors, adjusting switching thresholds and cutting the secondary loading pump's maximum flow rate by 39 % (11,500 to 7,000 l/h) lowers the modelled mean return temperature by 5.6 K to 47.9 °C, without structural work. Dedicating one buffer tank to hot water and the other to space heating gives 43.2 °C. The largest effect comes from the cascade: rather than supplying space heating and hot water in parallel, the primary water passes the hot water heat exchanger first and only then, in series, the space heating loop. The principle is known from substation practice; new is its quantification for an existing building with buffer tanks: 41.3 °C annual mean, 12.2 K below the reference, well beyond the 2 to roughly 10 K otherwise reported without a substation heat pump. It is paid for in comfort: for 190 hours a year the heating supply is 1 K below target, for 26 hours 3 K. Separating the buffer tanks gives four fifths of the benefit at around 10 K, with 22 and 3 hours.
The authors extrapolate to network level, assuming every connection was operated like this building. The spread would rise from 36.5 to 48.7 K, so only 75 % of the volume flow would be needed. Importantly, the gain can be spent only once: either pump electricity falls – the quoted 42 % of previous pump power is the theoretical affinity limit; with real differential pressure control 55 to 65 % is more realistic – or 33 % more heat is sold at the same flow rate, or the supply temperature comes down too, 78 °C instead of 90 °C at a constant 35 K spread, cutting heat losses by some 17 %. Peak power at the substation also drops from 240 to 140 kW, mainly through the limited loading pump: lower connected load, more capacity for densification without reinforcement.
Three caveats belong with this. None of it helps in summer: because of the hot water hygiene requirement – Austrian B1921 with 55 °C, stricter under German DVGW W 551 – the return stays at or above 53.9 °C in every variant. That is only 10 % of annual demand, but summer sets the minimum supply temperature; the authors propose a small air-source heat pump covering the circulation losses, 5 kW here. Second, this is a simulation feasibility study: calibrated against the existing system, the optimised variants not validated by measurement. Third, the cascade needs a high primary supply temperature and low-temperature-capable emitters; it is demonstrated for space heating to hot water ratios between 1.6 and 3.2, and the authors consider it unsuitable for passive houses. And every measure sits behind the substation: the assumed tariff saves the customer 1,141 € a year, too little to trigger a plant room conversion. The benefit accrues to the network.
The second paper looks at the whole network. In meshed networks with several suppliers and storages the flow directions are not known in advance; with temperature mixing at the nodes this yields a mixed-integer nonlinear, non-convex problem, usually solved heuristically. The authors solve it globally, with a stated optimality gap. The operational value lies less in the schedule than in the yardstick: the distance between today's operation and the optimum becomes quantifiable.
The most effective of five methodological ingredients rests on a simple physical insight: no circulating flow can exist around a cycle in the supply or return part, because the pressure differences sum to zero while the friction losses do not – so whole combinations of flow directions can be excluded in advance. Together with four further ingredients, the number of instances solved to optimality rises from 20 to 50 across 26 generated test networks in four load cases each, with running time down to some 40 %.
More telling are four real instances: aggregated models of two Darmstadt networks from ENTEGA Plus and TU Darmstadt, each in a present-day and a future variant with more decentralized suppliers. The present-day networks, with one and two suppliers, were solved within the 1 % gap (294 and 2,956 seconds). The future variants with six and nine suppliers were left after five hours, even in the best configuration, at gaps of 8 % and over 400 % – even though they are aggregated more coarsely and are therefore smaller. Difficulty is driven not by network size but by the number of feed-in points. Note the scope: the model is steady-state, capturing neither transport delays nor the network's own storage effect, the storages have no capacity limits or losses, and the customer return temperature is fixed at 40 °C. The very value the first paper works hard to achieve is assumed as given by the second.
As always, we recommend reading both articles in full; both are freely available, though the Darmstadt paper is a preprint without peer review, with code and instances open at github.com/dopt-TUDa/heatnet. Both studies stop where daily operation begins, and at the same point: distribution. The first extrapolates one model building to a network in which every building behaves differently; the second assumes a blanket customer return. The operator questions are therefore: which substations are spoiling your spread, and what is the gain worth given your differential pressure control and pump curves – as pump electricity, free capacity, or a lower supply temperature? That is what we represent in the heatbeat Digital Twin: networks with their real load profiles, substation metering data, hydraulics, and actual operating temperatures.
We’re also looking forward to the 31st Dresden District Heating Colloquium in September, where we’ll have a booth and will be on hand to show you the latest developments in the heatbeat Digital Twin. In addition, ENERGIEregion Nürnberg e.V. is organizing a dialogue event on September 28 on the topic “ Thermal Networks—Digitalization in Practice ,” to which we’re happy to contribute.
The next issue of our newsletter will be published on 7 October 2026.
Your heatbeat team