Communal Heating and CO2 Air Source Heat Pumps

From Heatweb Wiki
Jump to: navigation, search

Design Copyright 2020 Thermal Integration Ltd.

This is an article dedicated to the very specific task of driving services to a block of flats using air source heat pumps and a heat network as follows:

  • 50 Properties
  • CO2 Air Source Heat Pumps as primary heat source
  • Multiple locations for heat pumps, with roof as main location
  • Space on ground floor for buffer storage and plant

Further to these contract requirements, we have added further requirements:

  • Pipes sized to guidance, based on water velocities between 0.5 and 1.5m/s.
  • Light commercial pump head requirements (180kPa)
  • Pipework layout minimises lengths of hot pipework
  • Advanced BMS using open control technology, providing contract free SCADA for life
  • Minimal property service requirements (every 5-10 years)
  • Full technical compliance with latest CIBSE CP1 Codes of Practice

Heat Pump Selection


The design is based on the use of The Q-Ton CO2 Heat Pumps from Mitsubishi Heavy Industries.

  • 30kW per heat pump
  • 3+ Seasonal COP
  • Output temperatures up to 90C
  • Refrigerant is Carbon Dioxide with GWP of only 1

When designing heat networks it is more efficient to size pipework for normal peak loads and increase network temperatures to get to absolute peak loads (rare events). When designing for heat pumps this would normally require topping up boilers, however the CO2 heat pumps can deliver 90C if needed, making it possible to weather compensate the network and keep pipe sizes down.

The following tables show the COP performance in the 60-65C range. It is particularly affected by the return temperature, with big gains from keeping the return below 30C. This is perfect for our industry leading HIUs which demonstrate consistent sub 30C network return temperatures in the field, and in the lab have the best independent return temperature results of any HIU in the industry at 15.8C over a year.



COP Analysis

From the COP tables the following can be seen:

  • The output temperature has least effect on the COP. In fact, you get a better COP at 65C in the target range (24C rtn) than at 60C.
  • The inlet (return) temperatures are most significant, with large gains to be had by pulling the return down below 30C.
  • The air temperatures also have a significant effect.

The first point lets us run the network at temperatures that have the most beneficial effects elsewhere.

The second point, lowering return temperatures, is down to HIU design and control strategy. For 3 years now we have had the industry's best BESA test figures for HIUs, with our DATA HIU having a VWART of 15.8C. In reality, mains water temperatures are higher than the tests, however return temperatures should be below 30C at nearly all times, even with heating (underfloor) turned on. Strategies to keep return temperatures low revolve around pipe sizing and keep warm approach.

Screenshot 2020-10-01 Heatweb MQTT Desk.png

The third point, air temperatures, leads to the strategy of using building extract air where possible/available.

The seasonal COP should be between 3 and 4 depending on how well heating systems are balanced and timed.

Electrical Supply

13kW Electrical Input per 30kW Q-Ton.



Copyright Thermal Integraton 2020.

Q-Ton Network 3.jpeg

Pipe Sizing

From CIPHE Building Engineering Services Design Guide:


Flow (hot) pipes are sized to achieve 1.5m/s (nearest pipe size up) at peak design load with primaries at 70/80C (0/-5). Although the system can achieve 90C at the limit, we are unable to feed higher temperatures through to underfloor heating so the benefits drop off.

Return pipes are sized to 0.5m/s (nearest pipe size down), as response times and thermal losses are not as important as pressure drops. i.e. The pump head is saved for the flow pipes which need to be fast responding.

Should electricity pricing make it practical, there is the option to bring storage up to 90C in advance of peaks in electricity costs.

Under normal conditions, the system will run between 52C and 60C.

The following chart uses a peak supply temperature of 85C for central heating and 60C for domestic hot water, with a DHW return temperature of 25C and a central heating return temperature of 30C. Domestic hot water diversity follows DS439. Central heating diversity has not been applied.

54mm Copper Pipe goes from 35 properties up to 70 properties, and also covers the peak combined heat pump input. With 54mm copper as the largest pipe size, the cost of pipework drops almost exponentially. £12.53 for a 54mm tee piece compared to £56.34 for 110mm for an end-feed tee piece - a reduction in cost of over 75% on size alone. Then one should consider the labour involved in making pipework connections in copper or plastic as opposed to welding steel (for example). Insulation and supporting costs all factor in, making the difference more marked. Costs should reflect the fact that pipework is (just) in domestic realms - once over 54mm you start to move away from domestic pricing and installation methods.


Pipe selection where multiple sizes apply between 0.5 and 1.5m/s should be based on:

  1. Flow pipes that can be 35mm or less should be sized to the smaller pipe size.
  2. Return pipes are sized to the larger pipe size.
  3. Remaining flow pipes should be smaller if pump head allows. If not then upsize the largest pipes first.

This is in recognition of the fact that pipes 35mm or less may stand idle for significaant periods, and heat losses drop off rapidly (see below). These pipes need to contain the smallest volumes as sizing allows so they can reheat as quickly as possible. Pipes larger than 35mm cool significantly slower, and in use may never cool enough for heat loss to significantly reduce.

Thermal bypasses are fitted no more than 10 litres of flow pipe volue upstream of the index properties.




Network circulating pumps to properties are located on the return side of the buffer store. This makes no change to the circulatory function, however it allows the hot primary supply to sit at neutral pressures on a common head with the hot feed from the buffer store as well as all heat sources. In turn this avoids the need to run separate hot pipework from the heat sources to the buffer reducing hot pipework significantly.

The design differential pressure is 1 bar (100kPa). 150/180kPa pumps are selected to achieve 100kPa at design flow. Pumps are Duty/Standby.

The Grundfos Magna 3 is selected for its ability to manage conditions of no-flow for extended periods. A Magna 3D 40-150 would be suitable for 50 properties.



Heat sources are assumed to have their own pumps, controlled to drive the output of the heat source, in this case 30kW per heat pump. Heat sources do not share the return pipe with distribution from properties, maintaining a dedicated feed pipe from the cold side of the buffer store. Minimal DP is generated by distribution across the heat sources.

Network Interconnection

The use of a common hot header for heat input and load, achieved by locating distribution pumps on the return side, allows networks to be more efficiently inter-connected, with the point of interconnection not tied to the plant-room.

Operational Functions

  • Network supply temperature linked to pump speed, with temperatures increasing as pump approaches upper limits.
  • Network wide heating over-ride, allows loads to be shifted away from (short) peaks in electricity pricing or domestic hot water demand.

The following graph is a handover demonstration with three heating over-ride activations.

Heating overide 2.png

Direct Heating

Direct central heating is prefferred for the following reasons:

  • Marginally lower return temperatures
  • One network to maintain water quality for, as opposed to 100s
  • Less equipment to install - a smaller HIU
  • No discharge pipes required

Efficiency (operational profit) is affected by heating distribution, so offloading water quality to end user is counter-productive. One large system can afford the best in protection and alarming, as so much rests on keeping systems clean. There is no reason why a clean, dosed, air free and pH balanced system cannot last for a many decades. The distrubution can be truly fit and forget with the correct alarming in place.

The biggest fear with direct systems is leaks - what happens when an end user undoes something they shouldn't, such as a radiator? The network detailed here has benefits that prevent the discharge of large volumes of water under failure conditions.

Firstly, return connections from loads are fitted with non-return valves, preventing water from discharging from the system via the return. This leaves the flow pipe. Pumps are fitted on the return, along with the network expansion vessel - compressed air in the vessel cannot contribute to driving flow. The flow pipe is continuously connected to the buffer store, which in turn also has an expansion vessel, however this is a smaller vessel, and flow in and out is restricted. The buffer, the flow pipe and the buffer vessel are the neutral point in the system. Assuming their is no excess air in the system, opening a connection on the hot districution would cause flow driven purely by the discharge from the buffer expansion vessel, limited to a low flow rate.

Advanced Temperature Drops

Where are the theoretical limits on efficiency in using a heat pump on a heat network?

Central Heating

Installing radiators in series with underfloor heating provides the follwing benefits:

  • The full temperature drop across the radiant system (rad+UFH) is maintained, with no loss of Entropy.
  • Return temperatures will be as low as possible, approaching room temperature.
  • Peak output becomes a function of network flow temperature, adjustable between 53C (for DHW at 50C) and how hot one can take the radiator (83C typically based on historic domestic gas boilers, although we can go to 90C with the CO2 heat pump). A floor to ceiling radiator would represent the limits of possibility, with 90C at high (safe) levels, dropping to 55C near the base, then into the UFH and out at 20C. 70C drop would be a theoretical limit - as some people have room stats higher than 19C.
  • COP of heat pump significantly improved by reducing return temperatures from heating loads. This will be most felt as load increases, with the flow temperature climbing while the return stays low.

This can be achieved using the following equipment for example:

Direct heating is preferable for a number of reasons, with installation and maintenance cost savings been the greatest driver.

Domestic Hot Water

While lower return temperatures are gained by having a colder mains water temperature, in energy terms, the hotter the mains the better. Less work to do in the first place whichever way you look at it, so thoughts to cool incoming cold using exhaust air are only for keeping cold at the right Legionella temperatures and no more.

So, all energy gains are to be had be eliminating uneccessary keep warm recirculation. Our HIU functions cover the control side to achieve this, however the limits are imposed by pipework layouts, minimising the lengths of lateral pipework. The best layout is with properties fed in back-to-back pairs off risers, with no laterals at all. The risers will generall be of a size that overnight heat loss is slow enough to prevent any keep warm flow, or a very small trickle.

At the very limits we are talking pipe sizing for peak DHW load. If one has multiple risers, connected at the top along with heat pump input, then also connected at the bottom, with buffer input, then pressure drops are lower and pipework sizes may be sqeezed down further potentially.

Cost Savings

The CO2 building heat network represents a step change in costs in all areas.

  • Pipework is dramatically reduced, both in length and diameter, pulling it into domestic realms.
  • The network is decentralised, with no need to build a plantroom. You need a buffer store, and a controls skid, however these are pre-assembled and require minimal floor space. The main driver for an expensive plant-room is the heat generators (normally boilers and flues).
  • BMS regarding the hydraulic side is no longer needed, with all network control (i.e. pump control and alarming) provided by the panel on the hydraulic skid, or heat pump controls.
  • Thermal losses are greatly reduced by keeping hot pipework to an absolute minimum, and locating all controls on return pipework at under 30C.
  • Modular 30kW approach to heat input works very well with HIU diversity and redundency, reducing overall capital costs. Additional loads can be met with additional distributed heat inputs and buffer stores.
  • Design charges and meetings. Schematic layout and pipe sizing varies very little. All controls logic is by manufacturers.
  • Apart from the heat pumps, the installation costs for the heat network change from a commercial installation to more of a join-the-dots pipework and cabling installation that any competent domestic installer can take on. With the largest pipe size been 54mm copper (or plastic), the need for specialist jointing methods is removed.
  • Commissioning costs are slashed, with kit setup by manufacturers.
  • The big one is running costs. The thermal efficiency of the network layout, combined with the COP achieved from partnering CO2 heat pumps with the latest generation of our HIUs, reduces the electrical consumption needed to drive the system.
  • Carbon off-setting costs - it possible achieve a zero carbon solutions by a combination of adding local PV panels as one can, to drive heat pumps, and pulling electricity from the grid. The latter is covered by carbon off-setting (for example funding a wind turbine somewhere windy). Cutting away the losses and boosting COP reduces the amount of carbon to off-set.

Parts List for 70 Properties

The following equipment can be supplied (not installed) with commisisoning included.

All components are designed to meet the very latest CP1 CIBSE Codes of Practice for Heat Networks, as well as pipe sizing guidance (CIPHE / CIBSE Guide B).

Quantity Description
7-8 Q-Ton Heat Pumps
70 DATA SPLIT (Indirect) or MIX (Direct) HIUs fitted with Heat Meters
10 HIU Communications Hubs (metering & HIU comms)
2 500 litre Buffer Stores (with sparges)
1 Network Hydraulics Skid (Pumps, Pressurisation, Side-Stream Filter, Sensors, Control Panel, Expansion Vessels)
3 Auto Air Vent assemblies (for high points)
2 Index Unit (DP Sensor & Thermal Bypass to go at distribution index points)



Advanced Electronic HIU, direct or indirect on same first fix rail, with industry leading technology as demonstrated in BESA tests.

HIU Communications Hubs

A Node-RED bridge between a group of up to 8 HIUs through to the Hydraulic Skid over Ethernet using MQTT protocols.

Network Hydraulic Skid

  • Node-RED Industrial Control Panel with
    • GSM Modem
    • PoE Ethernet Switch
    • USSD Backup Modem with 10 years connectivity
    • Redundent Power Supplies for Controls
    • MQTT Server
    • Logic Control
    • Touch Screen Display
    • Industrial I/O (ditial/analogue)
  • Network Pumps - Magna3D 40-150
  • Dosing Pot
  • Dirt & Air Seperator
  • Side-Stream Filter - ENWA EM825 (with Modbus)
  • Pressurisation Unit
  • Filling Loop
  • Expansion Vessels
  • Heat Meters - Zenner Celcius (M-Bus)
  • Sensors (temperature/pressure/dp)
  • Isolating Valves
  • 2" (54mm) Network Connections
  • Factory assembled, wired & tested

Expansion Vessels

Allowing up to 4500 litres of buffer, and a network at 500 litres, we would like 700 litre expansion vessels.

Selection is:

1 x 500 litre vessel on buffer stores / positive side of pumps. (3 bar precharge) 1 x 300 litre vessel on negative side of pump. (2.0 bar precharge)

Based on fill pressure of 3.5 bar at lowest point, along with allowance for wide temperature ranges, redundency, and differential pressure storage.

One vessel is fitted on the flow side of the pumps an one on the return. The larger vessel, with a higher precharge, acts as the neutral point. The different precharge pressures results in more water sitting in the return vessel, and more air in the main vessel. When the pump is running the levels will be more equal. Individual tap uses, typically less than 10s, are expected to be met by vessel pressure difference as they are too brief for pump response. The two vessels provide a flow dampening effect at low loads making pump control/response more like a DHW pressure set - to meet DHW loads.