11 replies to this topic

[portal266]

Hello all,

 

In a district heating network, pressure boosting stations are installed. Each station has a pump installed on the supply pipeline and a control valve installed on the return pipeline.
The boosting station is controlled in such a way that the pump regulates the available pressure so that, at the most distant customer, the available pressure does not drop below 80 kPa.

 

The control valve on the return line is used to regulate the upstream pressure.

 

My understanding is that when customers draw too much water, the control valve protects the pump from operating in inefficient regions of its characteristic curve (low head and high flow). By partially closing the valve, the pump is forced to operate at a higher pressure and lower flow, while still maintaining the minimum required available pressure at the end customer.

 

Can someone confirm this or point out any errors in my reasoning?

[Pilesar]

If it were my system assuming one pressure measurement at the most distant customer... I would have the pump run all the time. If the pressure were above 90 kPa, then open the control valve. Below 90 kPa, the control valve would be completely closed. Pump efficiency would not be regulated -- it would just be wherever the pump curve falls. A centrifugal pump will have a minimum flow requirement which may need to be incorporated in the control scheme.

[portal266]

Thanks for your response.

 

The pump is controlled by a VFD. This is a system supplying domestic hot water heat and central heating exchangers from a hot header (N consumers).

If the last consumer has low available differential pressure, it means the pump was not able to provide sufficient pressure to cover the losses resulting from high water demand. In that case, throttling the valve makes sense, because it steepens the system curve. This gives us the possibility to rebuild pressure at the expense of lower flow rate.

I think that without a return valve, the pump would indeed deliver a high flow, but it would not generate enough pressure for any flow to pass through the heat exchanger at the most distant consumers. Therefore, this is an intentional energy loss in situations where consumers receive too little energy from temperature alone and try to compensate by increasing flow. I think that’s the point of it.

[Pilesar]

I think that without a return valve, the pump would indeed deliver a high flow, but it would not generate enough pressure [...]

Centrifugal pump curves are generally such that at high discharge pressure, the flow is lower than at low discharge pressure for a given suction pressure. Is this a closed system? Or is there constant fresh makeup water to the header? Does the return go to a hotwell tank or straight to the suction of the pump? It sounds like you have an interesting system that is difficult to describe completely with words. Drawings would be helpful in communicating. I don't want to mislead you with my comments by addressing features of your system that do not exist.

[portal266]

This is a district heating network, which can be divided into three main parts: the heat producer, the heat distributor, and the customers (buildings, residential areas, factories, etc.).

 

The system is closed, meaning that, in principle, the amount of water leaving the heat source returns to it. If less water returns, the deficit is made up (I am not certain whether the makeup water is provided by the producer or the distributor).

 

I do not have detailed information about the internal configuration of the heat source itself, but the heat distributor must ensure that the return pressure to the heat source is maintained at approximately 200 kPa. The distributor orders water from the producer at a specified supply temperature, typically in the range of 70–120 °C. In the past, I asked whether the distributor could also request a specific supply pressure; this is possible and occasionally done, depending on the expected flow rates in the network (maximum flows can exceed 10000 t/h).

 

More than half of the city is supplied directly from the heat source through the distributor’s network. In areas with significant elevation differences, the distributor has installed pressure boosting stations. Some of these stations are connected in series, and heat consumers are located between them. The consumers are connected via heat exchangers installed in parallel between the supply and return headers. Downstream of the heat exchangers, control valves modulate based on the customer circuit water temperature; if the temperature is too low, the valves open.

 

Overall, the installation is indeed very complex and difficult to describe fully in words, so I am including a simplified schematic to better illustrate the process.

 

 

 

 

 

Edited by portal266, 29 January 2026 - 07:59 AM.

[Pilesar]

You wrote: "My understanding is that when customers draw too much water, the control valve protects the pump from operating in inefficient regions of its characteristic curve"

 

The diagram does not make me an expert, but I see nothing to support your understanding of the control valve purpose. The control valve on the return line is concerned with the upstream pressure only. The pump will deal with whatever suction pressure it gets. A pump designed for this system would have a large operating range. The VFD handles the demand. In theory, pumps should run at their best efficiency point. In real life, pumps run anywhere on their curve which could be in the 'barely efficient' region. Because the pump demand is so variable, the VFD helps keep the pump out of the 'terrible efficiency' zone.

 

Is your interest an academic exercise? Are you student? I have always found closed circulating systems to be non-intuitive and I am glad I never had to design one from scratch.

[portal266]

I may not have described my thinking clearly enough. I’m not coming at this from a pump design background; I’m a software developer trying to understand the idea behind this type of control.

Let me try to explain how I currently understand the sequence of events, using a simple example.

Assume the pump can operate from 20–100% speed and is currently running at about 50%, slightly to the right of its best efficiency point. The differential pressure at the end users is controlled to a bit above 80 kPa. 

 

When the weather gets colder, the customer valves open slightly, which increases the flow. As the flow increases, the control valve on the return side starts to close in order to prevent the pump from losing further differential pressure due to the higher flow.

 

At the same time, the higher flow through the network increases pressure losses in the pipes, and the available differential pressure at the most remote customers drops below 80 kPa. The VFD then increases pump speed, restoring the required differential pressure for those customers.

 

This topic interests me because I haven’t been able to get a clear explanation of this control concept from people I know on the process side. Recently I watched a “Process with Pat” video about designing pressure drops in piping systems, and it made me think about the return-pressure-based control. Since we effectively choose the pressure upstream of the control valve, it seems possible to define it as a function of flow (in practice it looks more like a band than a single curve, with a spread of around 100 kPa).

I’ve been going through some district heating literature and found piezometric diagrams (pressure profile) that explain how pressures are shaped in the network. What confused me is that older references usually assume constant flow networks, while in this case the flow varies significantly due to control on the customer side.

Edited by portal266, 29 January 2026 - 09:42 AM.

[Pilesar]

The additional explanation helps. I think your description is accurate in its effects, but perhaps not in the reasons for the control. I cannot say you are wrong -- just that I am suspicious. My understanding: The control valve controls pressure, not flow. There may be a correlation to flow, but this would be indirect. What if a substation branch were blocked in? The flow would be changed, but the control valve might not change since the resistance of the circuit would change. When modeling controls, it is best to use the same variable that the controller uses. I have developed inferential controllers in cases where the real variable of concern is not measured. These are used in 'advanced control schemes' that are almost impossible to tune manually but require much historical data. Because parameters change in the real world, the inferential measurement calculations must be checked periodically and updated for changed conditions. I bring this up because even if you could accurately model control valve position as a function of flow today, it may not be anywhere close to correct next week. 

   You don't say that you are building a software model of this system. In building accurate process models, some relationships like pressure drop can be fixed if the flow rate is fixed. With changing flows, pressure drop changes may be fundamental. I have built dynamic models for operator training and it is not easy to capture all the important parameters. If you were to ask several operators of your system how the controls work, you might get different answers! For the real answer, try to find the operating manual the original designer of the system should have developed. These manuals may not ever be read but often have insight into the thinking behind the system. You might also ask the shift supervisor of the system. That person should understand its functions. Even if it is not your business to know, they may appreciate the opportunity to show off their knowledge.

  You brought an interesting question. I cannot give you a definite answer but enjoyed thinking about it. The engineering field is made for the curious. There are many ways to use your software development skills in the process engineering world.

[portal266]

Thank you for your time and engagement

 

This is a very interesting topic, and I have been wondering for a long time how it actually works. My colleagues asked the operators how they set the setpoints on the differential pressure and return pressure controllers, but none of the operators were able to give a clear answer. They simply set them empirically. One operator may do it better than another, and the same applies to estimating the load and ordering the supply water temperature.

Only once did I hear that the return pressure helps maintain the correct differential pressure between the supply and return header, but beyond that I could not get any more detailed explanation.

[Pilesar]

I have designed process systems and provided written instructions. When put into the field, sometimes the features and great ideas I incorporated would never even be tried! I still believe it would work better if it were operated according to the way I designed it.

   On the other hand, I was once sent to another city to help start up a unique process someone else designed that operations could not start. I tried it according to the designer's instruction and it should have worked according to theory but did not. I evaluated the safety features and tried a completely different operation method which was much simpler. Everyone was surprised (including me) that the process ran flawlessly. Much of the expensive components were not even needed and just added complexity. 

   I've been spending many hours the past few weeks building PowerShell scripts. Trial and error is my main mode since I've had no training in the code. Microsoft CoPilot has been a huge assistant. The more I use CoPilot, the better I like it. I know it gives me wrong answers often (which are frustrating) but it is good at providing ideas I never would have considered. I bring this up just to say that discussing problems and getting other opinions and viewpoints provides better results than trying to figure out everything by yourself. I wish you success on your quest.

[portal266]

Thanks for sharing this. Discussing problems on a forum is very valuable and saves a lot of time. One person mentions something, another corrects or adds to it, and in this way the possible solution paths get narrower and sometimes someone provides the solution outright

 

Regarding the problem being discussed, yesterday I came across an article entitled “Reduce Pressure and Heat Losses Through Intelligent Utilisation of Pumps” (see Fig. 4), which presents the pressure profile of a system with a single pump at the source and distributed pumps in individual network sections. What doesn’t make sense to me is the return pressure profile. For water to always flow in the same direction, the pressure downstream of the valve (on the heat source side) must be lower than the pressure upstream of the valve (energy conservation principle).

 

I decided to check the pressure drop from the pump discharge to the most remote consumer - the supply pressure. I have several critical consumers, and I selected those with the lowest pressure difference across the heat exchanger. As expected, the pressure drops were positive, consistent with the pressure profile diagrams.

Then I checked the same for the return header. To my surprise, the pressure drops calculated from downstream of the valve at the heat exchanger to the pressure boosting station were both positive and negative. This is inconsistent with what is shown in the diagrams.

 

I concluded that the pressure downstream of the valve at the most remote consumer would be lower than the return pressure at the pressure boosting station in two cases.
The first case is when there is such a large heat deficit that everyone opens their valves downstream of the heat exchangers, causing the pressure in the pipeline to the end user to drop so much that the user receives no water at all - essentially trying to push water from an empty supply pipe into an empty return pipe.

The second case is when everything is operating correctly, meaning there is sufficient pressure at the supply side of the last consumer and the condition of more than 80 kPa is met, but that consumer’s valve is closed. Since it is the last one on the line, the water in the return pipeline drains, leaving the most remote section of the return line empty.

 

If the system actually behaves this way, then the control valve at the pressure boosting station should start closing when the pump is operating at maximum capacity and the pressure difference at the end consumer drops below 80 kPa. The effect of partially closing this valve in such a situation is to build up pressure upstream of the valve and restrict closer consumers, so that water can reach the most remote ones.

 

Just to be transparent - I also use AI to help translate my replies. I understand your posts without any problem, but I use a translator to make my own answers clearer and easier to follow

[Pilesar]

For the benefit of others, I attach the link to the article you mention: https://api.grundfos...ure-6512062.pdf The article discusses the situation you described. District heating is not something familiar to me and I do not want to make myself an expert on it. However, I found the article well-written and somewhat interesting as an introduction.