Regeneration is a well-known solution to the problem of imbalance in geothermal borefields. Designing these regeneration systems, however, is not trivial. In this article, we continue our discussion of designing borefields with regeneration in GHEtool by looking at a typical example: an extraction-dominated residential building with a dry cooler to counteract the imbalance.
!Note
This article builds on our previous article on regeneration. To understand the content below, we highly recommend reading that one first.
What is regeneration?
When designing shallow geothermal borefields, you often have to deal with long-term temperature drift caused by the imbalance of your system. Your borefield will either cool down year after year because the extraction load is higher than the injection load, or heat up when the injection load is dominant. In both cases, imbalance is undesirable, since it typically requires a larger borefield to keep the fluid temperatures within the set limits.
There are different ways of dealing with this imbalance: you can avoid it, opt for a hybrid system (taking part of the load away from the borefield) or choose a system with regeneration, where another technology is used to counteract the imbalance, reducing the long-term temperature drift and keeping the ground (more) balanced over time.
In our previous article on regeneration (which you can find here), we explained the design complexity of borefields with regeneration in detail and discussed different regeneration algorithms and technologies. In this article, we apply all that theoretical knowledge to a typical example of a residential building with an extraction-dominated borefield, to really understand the differences in design methodology.
!Note
Since the first release of this method, back in September 2026, we have already received quite a lot of positive feedback from our community. One of the main comments was that sometimes, an hourly regeneration potential is already known. For example, you have waste heat available with a given, known profile. This case did not yet fit into the four existing options (constant power, constant temperature, solar thermal and dry cooler), which is why we added a fifth option to upload your own hourly regeneration potential profile.
Example in GHEtool
As an example of the advantage of regeneration and the different simulation options, let us consider a residential building that has a significant demand for heating and domestic hot water as well as some high cooling peaks. The borefield that is used for this building has 40 boreholes, with a constant flow rate of 8 kg/s of pure water. Due to the use of water without any antifreeze, the temperature limits are quite strict: the average fluid temperature should stay above 5 °C to avoid local freezing in the heat pump’s evaporator, and below 16 °C to ensure passive or free cooling in summer.
When the borefield is simulated with a simulation period of 40 years, the following temperature profile is obtained.
In the temperature profile above, the imbalance is clearly visible. In the first couple of years, both heating and cooling can be provided without any problem, but in the long run the geothermal system will fail due to the low temperature and the use of water as heat transfer fluid. One solution could be to increase the borefield size or to work with antifreeze, but another one is to add a regeneration technology to compensate for the imbalance and make the system work with the existing size. This last option is the one of interest in this article.
In the example below, the regeneration will be done using a dry cooler, which is an air heat exchanger that can be quantified based on the power per unit of temperature difference between the outside air and the fluid. This relationship is the slope in the figure below. Based on this performance measure, let us now consider different design options for the regeneration.
!Hint
Although it is possible to run the regeneration algorithms with a simulation period of 40 years, this is not recommended due to the longer computation time. Ideally, you start by comparing different design options with a shorter simulation time (2–5 years) to see their effect and only run for a longer time when you are quite sure about the regeneration technology and size you have selected. That way, you can design regeneration technologies as quickly as possible.
Inlet and outlet
To start, let us consider a dry cooler with a specific heat exchange of 5 kW/K, a 0 °C pinch temperature, a simulation horizon of 4380 hours and the yearly regeneration algorithm. The remaining design choice is whether to place the dry cooler at the borefield inlet or at the outlet.
Depending on your particular project, one or the other can yield more regeneration. In fact, in order to increase the regeneration capacity of any technology that depends on the temperature (i.e. constant temperature, solar thermal and dry cooler regeneration technologies), you want to maximise the temperature difference between the fluid and the air to increase the regenerative power.
When the dry cooler is installed at the inlet of the borefield, the fluid temperatures during extraction (i.e. heating and domestic hot water) will be the coldest, but they’ll be the highest during cooling (see also our article on inlet and outlet temperatures). Therefore, if you have relatively warm winters and cool summers, this could give you the highest regenerative power.
However, if your winters are cold and summers warm, so that most of the regeneration should happen in summer, it is best to place the dry cooler at the outlet of the borefield, so that the fluid temperatures are the coolest in summer. The final choice is hence strongly dependent on your particular weather profile and building load. In the graph below, the regeneration load is shown for the inlet position.
When the dry cooler is positioned at the borefield inlet, the imbalance drops from 89 MWh/year towards extraction to only 29 MWh/year, due to the 60 MWh/year of regeneration. As can be seen in the graph above, the regeneration takes place throughout the year, and the warmest periods of the year are avoided so that the fluid temperature stays within the limits. The resulting temperature profile is shown below.
Both in the temperature profile above and in the regeneration graph, it is clear that the dry cooler cannot regenerate much in the first year (there is a big gap in the regeneration power). This is because the ground temperature is still too high in the first year to regenerate a lot, since that would compromise our maximum fluid temperature limit and hence the passive cooling potential in summer.
In the graph below, the regeneration power is shown for a position at the borefield outlet.
Although the effect is subtle, in the case above, the regeneration happens less in the winter and mid-season months and more in the summer and warmer periods. This is precisely due to the colder fluid temperatures in summer and warmer ones in winter. In this particular case, the regeneration for an outlet position is slightly higher, at 63 MWh/year.
!Note
In practice, a heat pump – regeneration – borefield configuration is typically more robust and easier to control than when the regeneration is in front of the heat pump. This is because the regeneration power can vary rapidly (especially when using solar thermal technologies), so when the heat pump is placed right after the regeneration, it should be able to cope with potentially high temperature swings. Whenever that is an issue, placing the regeneration at the borefield inlet can be a safer configuration.
Pinch temperature
In the simulation above, the pinch temperature was set to 0 °C, which leads to the theoretical maximum potential. The pinch temperature is defined as the smallest temperature difference between the two sides of a heat exchanger (in the case of a dry cooler, the outside air temperature and the fluid temperature) and is always higher than 0 °C, since there is always some temperature loss across the heat exchanger. Depending on your dry cooler configuration, this pinch temperature will be larger or smaller. With a realistic pinch temperature of 2 °C, the resulting regeneration graph is shown below (again, and from now on, for a position at the borefield inlet).
As can be seen, there is now less regeneration in total (only 43 MWh/year instead of 60 MWh/year) and almost no regeneration in winter. This is because, conceptually, you need to subtract the pinch temperature from the temperature difference between the air and the fluid, which reduces the regeneration potential.
Different sizes
Without any pinch temperature, the inlet position still left about 29 MWh/year of imbalance, so a logical step, if we want to completely remove the imbalance, is to increase the size of the regeneration technology. When the dry cooler is doubled in size to a 10 kW/K machine, the following regeneration load is obtained (for a 0 °C pinch temperature).
When the profile above is compared to the 5 kW/K case, it is clear that a 10 kW/K dry cooler gives more regeneration (72 MWh/year instead of 30 MWh/year). However, when the dry cooler size is doubled once again to 20 kW/K, the regeneration load only increases to 77 MWh/year, meaning that for a doubling of the dry cooler size, only a 6% increase in regeneration capacity is obtained. The regeneration graph for the 20 kW/K case is shown below.
Although the regeneration power is higher in this case (reaching almost 100 kW), the period in summer in which there can be no regeneration due to the maximum temperature threshold becomes longer. This counteracts the positive effect of the higher regeneration power and in the end yields a result that is rather similar to the 10 kW/K case. The temperature profile for the regeneration graph above is shown below.
The regeneration potential is clearly limited by the maximum average fluid temperature here, so increasing the regeneration capacity even more will not decrease the imbalance any further. However, we still have 12 MWh/year of extraction imbalance left. There are two ways to solve this issue: one is to shorten the simulation period and the other is to increase your maximum allowed fluid temperature.
Shorter simulation horizon
The simulation horizon, which was previously set to 4380 hours (exactly half a year), determines how far into the future the algorithm looks. Regenerating the ground in the current hour affects the hours, weeks and even months to come. Therefore, if we regenerate too much before summer, the ground temperature can rise too high, thereby destroying any passive cooling potential.
To quantify this effect, let us simulate with a simulation horizon of just 500 hours, which is less than a month. Doing so removes the entire imbalance (now regenerating the full 89 MWh/year) and gives the temperature profile below.
As can be seen, the maximum average fluid temperature limit is crossed, with a maximum fluid temperature of 16.4 °C. However, the temperature violation is only 1 Kh, meaning that it only occurs for 2.5 hours each year (a 0.4 K difference for 2.5 hours gives exactly 1 Kh of temperature violation), which is negligible in most cases. The resulting regeneration profile is shown below.
When the regeneration graph above is compared to the case with a simulation horizon of 4380 hours, it is clear that there is now much more regeneration in summer and that the period without regeneration has become shorter. This is precisely the effect of a shorter simulation horizon, which can also be interpreted as a less strict regeneration simulation.
Different regeneration algorithms
Another way of increasing the regeneration capability of the same 20 kW/K dry cooler is to increase the maximum allowed fluid temperature to 18 °C. This can be done, for example, by changing the emission system in your building so that the same summer comfort can be guaranteed with a higher fluid temperature. With this maximum allowed average fluid temperature of 18 °C, let us now consider the four different regeneration algorithms: yearly, simulation period, proceeding horizon and proceeding horizon +.
Yearly
As explained in our previous article, the goal of the yearly regeneration algorithm is to reduce the imbalance in each year individually. The resulting regeneration profile is shown below.
In the regeneration graph above, it is clear that, due to the higher allowed fluid temperature, the regeneration can now also occur during the summer itself. Only one high cooling peak is avoided in order to stay below 18 °C. What is interesting, however, is that the regeneration apparently stops in September. This implies that, with the 20 kW/K dry cooler, the entire 89 MWh/year of regeneration has already happened by September, leaving us with a perfectly balanced borefield. The corresponding temperature profile is shown below.
Simulation period
Instead of looking at the imbalance from year to year, it is also possible to look at the cumulative imbalance for the entire simulation period. That is exactly what the ‘simulation period’ algorithm does. The resulting regeneration graph is shown below.
When the simulation period algorithm is selected, the same 89 MWh/year (on average) of regeneration can be obtained, but in a slightly different way. As can be seen in the profile above, the regeneration does not stop in September. Instead, the dry cooler keeps injecting heat into the ground until, sometime before the summer of the fifth year, the entire imbalance has been compensated for and regeneration is no longer needed. This gives the temperature profile below.
In the temperature profile above, the difference from the yearly algorithm is clearly visible in the last year. Here, the temperature profile shows no regeneration from April of the fifth year onwards, leading to much lower cooling temperatures and, for that particular year, an extraction-dominated imbalance. However, in the first four years, there is much more regeneration, as shown by the higher fluid temperatures in autumn compared with the yearly algorithm.
!Note
The simulation period algorithm is the best way to show what a specific regeneration technology is capable of, since all the other methods are limited either by the yearly imbalance or by the proceeding imbalance (as will be discussed in the next two subsections).
Proceeding horizon
The two algorithms above have the advantage that they can anticipate the imbalance, in the current year and in the entire simulation period respectively. That way, they can maximise their regeneration capabilities. In reality, however, we do not have such perfect information about the future imbalance. Therefore, a proceeding horizon methodology can be used, which only regenerates the imbalance that has already occurred up to the current hour. The regeneration graph is shown below.
The proceeding horizon algorithm has yet another shape, and what is interesting is that, in the first year, the regeneration now stops in July instead of September, as with the yearly algorithm. This is because in July, the cumulative imbalance from January up to that point has been fully compensated, and since the algorithm cannot anticipate future imbalance, it stops there. Only in winter will it start regenerating again, since there is once more some extraction load.
However, this approach means that we are always reacting to the imbalance instead of anticipating it, which means that we only regenerate 83 MWh/year on average, leaving an average imbalance of 6 MWh/year. The corresponding temperature profile is shown below, where the point at which the regeneration stops is clearly visible.
Proceeding horizon +
As a last option, the proceeding horizon + algorithm will be discussed. As the name suggests, this is an extension of the proceeding horizon algorithm, but with the option of regenerating in both injection and extraction. As mentioned above, the proceeding horizon algorithm stopped in summer because the cumulative imbalance was completely regenerated.
However, in July and August there was also an injection-dominated imbalance that could be compensated for. The proceeding horizon algorithm, however, only regenerates in the direction of the general imbalance (in this example, extraction-dominated), so it ignores the ‘local imbalance’ in summer. The proceeding horizon + algorithm does not, which results in the regeneration graph below.
The graph above clearly shows that, whereas the proceeding horizon algorithm simply stopped regenerating, the proceeding horizon + approach switches to extracting heat from the ground after summer, giving us both extraction and injection regeneration in this case. Interestingly, however, the resulting imbalance is exactly the same in both cases (proceeding horizon and proceeding horizon +): 6 MWh/year. The corresponding temperature profile is given below.
In the temperature profile above, a difference in fluid temperature between the proceeding horizon and proceeding horizon + algorithms is visible in autumn, where the latter lowers the fluid temperature slightly.
The advantage of the proceeding horizon + algorithm over the standard proceeding horizon algorithm is that the fluid temperatures are more levelled, since local imbalances are also accounted for. This can potentially lead to a higher efficiency in heating and cooling.
!Note
The advantage of the proceeding horizon + algorithm over the standard proceeding horizon algorithm is that the fluid temperatures are more levelled, since local imbalances are also accounted for. This can potentially lead to a higher efficiency in heating and cooling.Conclusion
In this article, we took a deep dive into simulating geothermal borefields with regeneration in GHEtool Cloud. Based on the example of a residential building with a strongly extraction-dominated load and a dry cooler to regenerate the ground, we discussed the effects of the placement of the regeneration technology, the pinch temperature, the simulation horizon and the different regeneration algorithms.
As with most things in life, and especially in geothermal engineering, there is no silver bullet when it comes to regeneration. Depending on your project, a different regeneration algorithm or different settings may work better. The key is to experiment and get a feel for how a specific regeneration approach behaves in your particular case.
References
- Watch our video explanation over on our YouTube page by clicking here.