the ratio of 𝑄
and 𝑄
, as only the latter changes
by the retrofit.
2.2 Appliances
Two types of air-source heat pumps were considered
in the calculations, as their application comes with a
lower limit in terms of the source side, compared to
ground-source or water-source heat pumps. One type
is the air-to-water heat pump that itself can cover both
heating and DHW net energy needs.
The other air-source heat pump type is air-to-air
heat pump. For DHW generation electric boilers are
assumed. As a reference, emissions of gas condensing
boiler systems are calculated as well. Carbon
footprint of the listed systems are calculated in the
followings.
2.2.1 TEWI of Heat Pump Systems
For the heat pump systems, TEWI calculations are
used to approximate the greenhouse gas emissions for
the lifetime of the appliance. TEWI considers two
types of emissions. Direct emissions are related to the
refrigerant used in the heat pump (described with the
first two terms of equation 1), while indirect
emissions are corresponding to the electricity used by
the heat pump (described with the third term) (Mota-
Babiloni et al., 2020).
𝑇𝐸𝑊𝐼
𝐺𝑊𝑃 ∗ 𝑚 ∗ 𝐿
∗𝑛
𝐺𝑊𝑃 ∗ 𝑚 ∗
1𝛼
𝐸
∗𝛽
∗𝑛
(1)
In TEWI, 𝐺𝑊𝑃 stands for the Global Warming
Potential (kgCO
2
eq./kg), 𝑚 (kg) represents the mass
of the refrigerant charge. 𝐿
(% / year) is the
annual loss of refrigerant charge through leakages and
𝛼
(%) stands for the amount of charge that can be
potentially recovered at the end of life. Factor 𝑛
(years) appearing is the estimated lifespan of the heat
pump.
The indirect emissions part is characterised with
𝐸
(kWh
e
/year), annual electricity consumption,
multiplied with the lifespan and applying the carbon
intensity factor 𝛽
(kgCO
2
eq./kWh
e
) of the
electricity, that describes the greenhouse gas
emissions that the generation of 1 kWh
e
of power is
accounted for.
2.2.2 Parameters of TEWI Calculation
In practice, TEWI is calculated for the heat pump that
is subject of the specific calculation. However, to
draw conclusions about the emissions of the different
heat pump-based systems under various conditions,
generalization of these values is necessary. In case of
the refrigerant parameters, and therefore the GWP
value, it is reasonable to define a type of refrigerant
for the study that is widely used. For that reason,
refrigerant R32 is taken into consideration, with a
global warming potential of 675 kgCO2eq./kg (Choi
et al., 2017).
The mass of the charge, though, is also dependant
on the specification of the heat pump and is in
accordance with the capacity of the appliance.
Usually, the specific charge mass is between 0.24 – 1
kg/kW (Poggi et al., 2008). In the study, 0.3 kg/kW is
used, just as applied in (Johnson, 2011). Leakage
rates vary on a wide scale in different studies,
however it is believed that they have a modest effect
on the results (Johnson, 2011). In case of the annual
leakage rate the value is usually around 6% (Greening
& Azapagic, 2012; Johnson, 2011). For recovery rate
also many different values can be found in literature.
Calculations of this paper assume 80% of recovery
rate, which is quite general (Greening & Azapagic,
2012). The expected lifespan of the appliances in the
following calculations is 15 year (Mota-Babiloni et
al., 2020).
Carbon intensity factor in TEWI is an average
value for the electricity used throughout the lifetime.
However, when calculating in advance, only
estimations could be used. Though carbon intensity is
believed to continuously decrease in the future, in this
study, carbon intensity of the installation year is used
through the lifetime, to rather err on the side of
caution.
2.2.3 Energy Consumption of the Heat
Pumps
It is often highlighted that the dominating part of
greenhouse gas emissions is the indirect part of
TEWI, energy consumption related parameters
(Greening & Azapagic, 2012; Mota-Babiloni et al.,
2020). Annual electricity consumption of this term
can be determined various ways. The most precise
would be measuring the consumption, however the
carbon emission estimation is carried out in advance,
therefore it is not possible in our case. Another option
could be a detailed dynamic modelling of the heat
pumps’ performance for the specific cases, which is
resource and time demanding. A simpler, though for