COP is used wrong in cooling for both Households and Buildings

[lodykuling]

The definitions for heating and cooling COP's are as follows (wikipedia): When one is interested in how well a machine cools, the COP is the ratio of the heat removed from the cold reservoir to input work. However, for heating, the COP is the ratio to input work of the heat removed from the cold reservoir plus the heat added to the hot reservoir by the input work.

Currently the COP definition for heating is used for cooling as well which leads to an underestimation of 1/3 of the energy needed for cooling. Fixing this problem will have quite a big impact on charts (as their values will change and their format will be outdated) and I guess on many other things as well, but I cannot oversee the entire impact this will have. Nor have I got a clue where is the best possibility for a quick fix.

The problem of using the wrong COP:

The COP for airconditioning as found in InputExcel --> cost heat pumps --> row K:

Useful demand as found in the graph:

[ChaelKruip]

@lodykuling please be more verbose. Now I am left with many questions...

• What do you mean with the red cross in the chart?
• Why do you point to that row of the Input Excel? What does it tell you?
• Why do you show the 'graph'?
• How are these numbers inconsistent?
• Which numbers could be correct?

I can guess/know some of the answers to these questions but that is not the point: it should be painfully/crystal/mediterranean sea clear to everyone (including your future self) what this ticket says. :smiley:

[lodykuling]

Ok, I will improve this ticket first thing tomorrow

[lodykuling]

The definitions for heating and cooling COP's are as follows (wikipedia): "When one is interested in how well a machine cools, the COP is the ratio of the heat removed from the cold reservoir to input work. However, for heating, the COP is the ratio to input work of the heat removed from the cold reservoir plus the heat added to the hot reservoir by the input work."

Currently the COP definition for heating is used for cooling as well which either leads to an underestimation of 1/3 of the final demand needed for cooling or an overestimation of useful demand by 1/4, depending on which way this is calculated. Fixing this problem will have quite a big impact, but I cannot oversee it entirely.

To reproduce this issue take the following steps with the COP definitions in mind (the graphical material is an example for airconditioning in households in a blankscenario):

1) Check useful demand for cooling in the following converters: buildings_useful_demand_after_insulation_recirculation_recovery_cooling or households_useful_demand_for_cooling_after_insulation

2) Check chart 66 for households and 71 for buildings. You will see that the sum of the electricity used and the ambient cold is equal to the useful demand found in step 1. This is the right way for heating COP, but wrong for the cooling COP.

3) Check the COP listed in (one of) the following converters: households_cooling_airconditioning_electricity
households_cooling_heatpump_ground_water_electricity
households_cooling_heatpump_air_water_electricity buildings_cooling_heatpump_air_water_network_gas buildings_cooling_airconditioning_electricity buildings_cooling_collective_heatpump_water_water_ts_electricity

They are respectively: 4, 4.6, 4.5 ,1.7, 4 and 10

4) Check the COP's in inputexcel they can be found in the tab called Cost heat pumps, spefically in row 77. They are respectively: 3, 3.6, 3.5 ,0.7, 3 and 9

So all COP's for cooling are increased by 1 before set into a converter. To fix this problem there are 3 solutions I can think of, of which A is a quick but horrible fix and for B I don't really know how to implement it, but it would be the "right" one.

A) Decrease all cooling COP's in inputexcel by 1. This will give the right results and keeps charts intact. But in the long run this will not be maintainable as updating with new efficiencies will likely not be done with this in mind. Besides, while the chart structure of No. 66 and 71 can stay the same their format still indicates that we're using the wrong method.

B) Make sure that the value set in the converters are not the COP+1, but just the COP for all cooling options. Besides that the split in the coverter which determines the percentage of electricity and the percentage of ambient cold should be changed to the new COP. Concerning the charts there are 2 options of which the second one has my personal preference: -Charts No. 66 and 71 can stay the same their format still indicates that we're using the wrong method.
-Charts No. 66 and 71 have to be redesigned.

[ChaelKruip]

@lodykuling thanks, this is much better. Now we should work on brevity a bit :wink:

[lodykuling]

I have traced this issue to the InputExcel in the tab 'Efficiencies'. Specifically the cells D94:D96 for Households and the cells D150:D152 for Buildings. (as an example I will continue with airconditioning for households)

To determine the fraction of work-input relative to the input of ambient cold, the following calculation takes place in these cells: =1/(1+HLOOKUP(A96,'Cost heat pumps'!$1:$85,58,0)) Which is effectively: =1/(1+COP)

However as the definition of COP= Q/W, on can determine the percentual input of W relative to Q by W/Q=1/COP or =1/(HLOOKUP(A96,'Cost heat pumps'!$1:$85,58,0))

Fixing this would resolve most of this this ticket. However the indication that we use the wrong method by showing charts 66 and 71 remains valid.

For heating with heat pumps there are no problems.

@wmeyers, @AlexanderWirtz do you agree this is wrong and that it has to be changed?

@ChaelKruip can you correct this problem as soon as WM and AW give the go sign?

[ChaelKruip]

@AlexanderWirtz what is the status here?

[AlexanderWirtz]

@ChaelKruip let's get together and fix this. Don't know where to do it now and would like to see current implementation in the code. Expect this is not an etmodel issue as far als @lodykuling 's last comment is concerned. Not sure I understand what the chart issue is

[ChaelKruip]

Assigning to @jorisberkhout

[jorisberkhout]

@AlexanderWirtz , as you are looking into the specs of heat pumps, could you advise me on the COP of heat pumps when it comes to cooling (both numbers and definitions)?

The ETDataset, ETSource and ETEngine code has changed significantly since @lodykuling wrote this issue, so I propose to start from the data and make sure that this ends up correctly in ETSource.

[github-actions[bot]]

This issue has had no activity for 60 days and will be closed in 7 days. Removing the "Stale" label or posting a comment will prevent it from being closed automatically. You can also add the "Pinned" label to ensure it isn't marked as stale in the future.

[ChaelKruip]

This has been nagging at me for a long while as I am not sure if there is a real problem. @mabijkerk could you look into this and see if it is still an issue? Refer to @michieldenhaan with your questions. If needed, we can discuss this in a broader group.

[github-actions[bot]]

This issue has had no activity for 60 days and will be closed in 7 days. Removing the "Stale" label or posting a comment will prevent it from being closed automatically. You can also add the "Pinned" label to ensure it isn't marked as stale in the future.

[mabijkerk]

Theory for documentation/background The diagram below shows the heating and cooling mode for a sample heat pump (air):

The difference is that in heating mode, the house is the heat sink (the heat needs to go to the house), while in cooling mode the house is the heat source (the heat needs to be removed from the house). In heating mode this means that the input electricity W_net adds useful heat as it increases the heat supplied to the house. In cooling mode the input electricity does not add useful heat, because it increases the heat supplied to the air, not the heat extracted from the house. Because COP = Q_useful / W_net applies for both heating and cooling mode, the COP for this sample heat pump is higher in heating mode than in cooling mode.

The issue in the ETM The problem now is that in the ETM, for example for the nodehouseholds_cooling_heatpump_air_water_electricity, the input electricity W_net does contribute to the cooling output. This is visible in the screenshot below, for a blank nl2019 scenario. The electricity input and cooling output are based on source_analyses in ETDataset, so the result is that the ambient cold input is underestimated.

The solution A straightforward way to solve this, is to add a loss, which is should be equal to the input electricity because the input electricity does not contribute to the useful demand for cooling. This approach requires minimal changes to the graph. In the example for nl2019 above the loss would become 0.31415 PJ and the ambient cold 1.41 PJ.

It is important to note that this means we assume the ambient cold extracted from the air to be equal to the cooling supplied to the house. This means that ambient cold extracted does not equal the heat that is actually supplied to the air, which would be Q_hot in the diagram for the sample heat pump in cooling mode. The distinction is that the ETM uses cooling, while in reality it is heat with a lower temperature.

To do

• [ ] Update cooling heat pump nodes to include a loss which is equal to the input electricity
• [ ] Update input electricity and input ambient cooling to be in line with this new approach. The node source analysis for these inputs seems to be quite outdated and lacks documentation, therefore it is better to research new data from the ECW and energy.nl (this can be done in parallel with https://github.com/quintel/etdataset/issues/900)
• [ ] Review charts that show cooling
• [ ] Review the calculation of the COP in the "Technical and financial properties" (this is calculated here https://github.com/quintel/etengine/blob/master/app/models/qernel/node_api/capacity_production.rb#L267-L280 where efficiency_based_on is the primary carrier, typically electricity and efficiency_balanced_with is typically ambient heat.)
• [ ] Write documentation for heat pump cooling