Content uploaded by Lai Fong Chiu
Author content
All content in this area was uploaded by Lai Fong Chiu on Apr 14, 2021
Content may be subject to copyright.
Chapter 60
Ecology of Heat Pump Performance:
A Socio-technical Analysis
Lai Fong Chiu and Robert Lowe
Abstract The UK government’s heat strategy is to reduce emissions from buildings
“to virtually zero by 2050” through a combination of technologies including heat
pumps (HPs). As part of this strategy, it introduced the Renewable Heat Premium
Payment (RHPP) scheme to incentivise the installation of HPs in the residential
sector. Using a socio-technical approach and case study method developed by the
authors in the field of energy research and building, this paper explores the rea-
sons for variation in performance of HPs supported by this scheme. Twenty-one
sites/households were selected for investigation. Owing to limited space, this paper
does not seek to present all cases, but instead focuses on key insights from five
cases that were originally thought to perform poorly. The findings highlight how the
complex ecology of a socio-technical system in determines performance. We will
show that system performance emerges from the dynamic interaction of monitoring
system, heat pump system configuration and occupants’ heating practices, and heat-
ing load factor. Limitations, practical implications, and scope for future research are
briefly discussed.
60.1 Introduction
The UK domestic sector accounts for approximately 29% of final energy consump-
tion, with space and hot water heat accounting for approximately 80% of this.
Enshrined in the sectoral plan of the government’s Carbon Plan [1] and Heat Strat-
egy [2] is the goal of reducing emissions from this sector to close to zero through a
combination of improved energy efficiency and deployment of low-carbon heating.
Heat pumps are seen as a key technology for achieving this, and the UK government
set up the Domestic Renewable Heat Incentive (RHI) scheme and the Renewable
Heat Premium Payment (RHPP) to promote the installation of HPs in this sector [3].
In parallel, it set up a field trial to understand the performance of HPs installed under
the RHPP scheme. This trial collected data for 700 of the c. 14,000 HPs installed
L. F. Chiu (B)·R. Lowe
University College London, London WC1H 0NN, UK
e-mail: laifong.chiu@ucl.ac.uk
© Springer Nature Singapore Pte Ltd. 2020
J. Littlewood et al. (eds.), Sustainability in Energy and Buildings, Smart Innovation,
Systems and Technologies 163, https://doi.org/10.1007/978-981-32- 9868-2_60
711
712 L. F. Chiu and R. Lowe
under the scheme, forming the largest HP field trial undertaken in the UK to-date.
Performance in the field is critical in terms of the economic competitiveness of the
technology, its acceptability to dwelling owners and occupants, and to enable the UK
government to demonstrate compliance with the UK’s target for renewable heat under
the EU Renewable Heat Directive [4]. Currently, the performance of HPs is defined
by two indices, the coefficient of performance (COP) and the seasonal performance
factor (SPF). The Renewable Heat Incentive scheme regulations [5] require that both
air source heat pumps (ASHPs) and ground source heat pumps (GSHPs) should
attain COP =2.9 and SPF =2.5 at boundary H4, to qualify for support under the
scheme. A team led by the second author of this paper was commissioned to analyse
remote monitoring data from the field trial and to undertake detailed case studies of a
sample of 21 installations using a socio-technical approach with the aim of improv-
ing understanding of remote monitoring systems, quality of metadata, reasons for
variations in performance, and occupant satisfaction with their HP systems. Owing
to limited space, this paper does not seek to present all cases, but instead focuses on
key insights from five cases that were originally thought to perform poorly.
60.2 Research Design and Methods
The present authors’ previous research in building performance has shown the neces-
sity of taking a socio-technical system perspective to empirical investigation of the
performance of energy technologies in the field [6]. This has resulted in the devel-
opment of a socio-technical case study method to collect and integrate data from
people and technical systems [7] that are interconnected, mutually adaptive, and
co-constituted. Below is a brief outline of the design and methods used for data col-
lection, analysis, and interpretation. A full report on the RHPP field trial case studies
is available online [8].
60.2.1 Sampling Strategy and Collection of Data
Based on statistical analysis of remote monitoring data, 117 householders and 31
registered social landlords (RSLs), representing a range of HP performances and geo-
graphic locations across the UK, were identified for initial contact. Positive responses
were received from approximately one-third of the householders, and almost half of
the tenants contacted by RSLs. Following a further round of selection, occupants of
21 dwellings, 7 under the ownership of RSLs and 14 owner-occupied, were recruited
to take part in case studies. Site visits of 2–3 h duration were made by a team that
included one technical researcher and at least one social researcher, who together
recorded detailed characteristics of each dwelling, configuration, and arrangements
of HP systems and interviewed occupants about their life-style, perceived thermal
60 Ecology of Heat Pump Performance: A Socio-technical Analysis 713
comfort in relation to the installation and operational experience of and satisfaction
with the HP during a room-by-room “walk-through” of the dwelling [8: 42].
60.2.2 Analysis
In the first instance, both quantitative and qualitative data were entered into a master
matrix. Analytic matrices were then constructed [8: App. 1] based on an investigative
logic of HP performance. This analytic framework gave primacy to thermodynamic
constraints, expressed qualitatively by the equation:
COP ≈0.5×330/T
(where T is the difference between the HP’s source and output temperatures), to
the configuration of system components and their interactions with built-form and
heat loss taken from Energy Performance Certificates (EPCs), to impacts of human
behaviours such as commissioning of systems by installers or occupants themselves,
and to occupants’ operating strategies in the context of lifestyles and costs.
Table 60.1 summarises the 16 cases in terms of types of HPs and their per-
formances. Analysis was initially performed on a sub-sample of 10 cases selected
from the 21 cases—these are arranged horizontally in Table 60.1. Because this sub-
sample included only a single RSL dwelling, the sub-sample was then expanded to
include all 7 RSL dwellings in the 21, bringing the total number of cases subjected to
detailed comparative analysis to 16. The additional six cases are arranged vertically
in Table 60.1. Estimates of SPFs for all systems were refined during the analysis.
Table 60.1 Cases classified by heat pump type, SPF band, and tenure
“well performing” heat pumps (SPF > 2.5)
“poorly performing” heat pumps (SPF < 2.5)
Heat pumps reclassified as “well performing” following analysis
RSL-owned dwellings shown in italics.
“AS” denotes air source and “GS” denotes ground source heat pump.
CS02
AS
CS03
GS
CS04
GS
CS05
GS
CS06
AS
CS18
GS
CS13
GS
CS19
AS
CS14
GS
CS20
AS
CS16
GS
CS12
GS
CS15
GS
CS09
AS
CS07
AS
CS08
AS
714 L. F. Chiu and R. Lowe
60.3 Factors Influencing Performance
60.3.1 Monitoring Systems and Data Quality
Seven “poorly performing cases”, CS02, CS03, CS07, CS09, CS12, CS15, CS16,
were initially suspected by the quantitative team of having monitoring system issues.
A combination of revised algorithms for automatic selection of monitoring data and
analysis of case study data led to SPFs for CS07, CS09, and CS15 being revised
upward to above SPF =2.5, leaving only four poorly performing cases.
Metering issues were one of a number of problems that affected CS07, including
a report from the occupants that their HP had suffered a “blow out” which was
corroborated by inspection of the monitoring data. These problems appeared to be
resolved following replacement of the faulty external unit. Monitoring issues were
also suspected in CS09; monitoring data indicated that the HP exhibited a persistent
but unexplained reduction in mass flow in the primary heating circuit as recorded by
the heat meter flow sensor. However, in the interview, the occupants reported that
there had been a flat battery in the monitoring system in the initial period following
installation which, once detected, had been replaced. The SPF for this system was
re-estimated based on data that was deemed not to have been affected, resulting in
reclassification as well as performance. CS15 had a monitoring issue due to faulty
installation of equipment, but this was corrected after its discovery.
When sensors were placed, [the] fitting was wrong – it was running backwards for a long
time – then they came back and altered it; they’d put the sensors on the wrong pipes; this
happened [maybe] around 18 months ago.
CS12 was suspected to have potentially serious and unresolved problems with
the monitoring system, which were confirmed during the site visit (see Sect. 3.2).
Metering issues were suspected in CS16 due to a low SPF, but subsequent analysis
suggested that this was better explained by the low load factor of the heating system
(see Sect. 3.3), which in turn arose from the occupants’ heating practices. Finally,
metering issues were also suspected in CS13 and CS18, due to data showing heat
output with no electricity input for short periods.
60.3.2 System Configuration and Heating Practices
The five well performing cases, CS13, CS14, CS18, CS19, and CS20, with SPF well
above 3 seemed to have well-insulated systems. Well-insulated internal pipework was
also found in two of the reclassified cases, CS15 and CS09, suggesting an association
with performance.
Metering data for CS09 indicated the presence of resistance heating for domestic
hot water. This appears to correspond to CS09’s everyday heating practices. As a
couple of retirees, the occupants lived mostly in the kitchen during the day-time,
60 Ecology of Heat Pump Performance: A Socio-technical Analysis 715
retreating in the evening into a small study to watch television; hence the living
room was rarely used. The temperature in the kitchen tended to be set at 18 °C, as
“movement [of] the sun & heat from cooking [would] make it comfortable [enough]”.
The kitchen also housed a large electric oven (an Aga). It is possible that the occupants
had been using the electric Aga to provide hot water but were not aware that this
might be costly.
Believing that they could keep themselves comfortable with lower costs, occu-
pants of CS09 controlled the temperature in the dwelling by switching their HP on
and off and by altering its flow temperature. They said that they did not find the
temperature so controllable using the thermostat. The researchers observed that the
thermostat was sited in an unheated lobby with a door often closed between the lobby
and the living space. This might have caused a higher than set-point temperature in
the rest of the dwelling, prompting the occupants to adopt the on–off control strategy
to keep the temperature down. From the documentation, less-than-adequate airtight-
ness of the dwelling was discovered by pressurisation test at the time of installation
of the HP. This might have added to the poor performance recorded at the outset.
CS12 was another poorly performing case with SPF initially estimated as 0.7.
The heating system was characterised by un-insulated pipework (see Fig. 60.1).
CS12 was a GSHP. Its hot water cylinder was housed in a cupboard within a heated
utility room. There was underfloor heating installed on the ground floor but radiators
upstairs. Initially, it was thought that internal uninsulated pipework should not affect
Fig. 60.1 CS12—uninsulated pipework associated with heat pump
716 L. F. Chiu and R. Lowe
performance. However, CS12’s room thermostat was in the kitchen immediately
adjacent to the heated utility room. The room thermostat in the kitchen could have
been affected both by heat gain from cooking and uninsulated pipes in the adjacent
utility space. Either factor might in turn have increased the tendency of this system
to cycle.
Monitoring data showed that heat demand for space and water heating was below
electricity input in every month for two consecutive years. An inquiry was made into
whether the data could be subject to metering error. Two features of the monthly mean
electricity and heat demand plot suggest it might have been. First, the extreme season-
ality of hot water demand and the persistence of space heating demand through the
summer suggested that there may have been a misallocation of total HP heat output
between space and water heating—this appeared likely in the light of a photograph
taken on site of the monitoring installation (Fig. 60.2), showing an auxiliary temper-
ature sensor on one of the primary circulation pipes, which had been installed with
insulation between it and the pipe. This could explain why the level of both space
and water heating recorded was implausibly low, even in a well-insulated dwelling
with a single occupant. The total heat output of the HP in January was sufficient to
raise the internal temperature of the dwelling by about 4 °C. Without other sources
Fig. 60.2 CS12—pipework showinginsulation between temperature sensor and pipe. The auxiliary
temperature sensor for the heat meter is just right-of-centre in a pocket. Source BRE
60 Ecology of Heat Pump Performance: A Socio-technical Analysis 717
of heat, this would have brought the mean internal temperature in January to around
8 °C. Solar gain in January is negligible, and internal free heat gains would have
been unlikely to add more than a few degrees centigrade to the internal temperature.
The site visit took place in December. Researchers observed that set-point tem-
peratures of most room thermostats in this dwelling were in the range of 17–18 °C.
Internal temperatures were not recorded, and it was therefore difficult to determine
whether set-point temperatures were reached or not. But the occupant stated that she
was “comfortable”. The manufacturer’s documentation showed the system installed
in this dwelling was fitted with either a 6 or a 9 kW “additional electric heater”. Given
doubts about the monitoring system, it was not possible to determine from the data
collected what proportion of HP output was produced by electric resistance heating.
However, plotting heat output from the HP against electricity input from available
physical monitoring data showed that a proportion of the heat output of this system
was derived from resistance heating, and that a proportion of the electricity input
resulted in no measured heat output. This is consistent with a statement made by the
occupant during interview that the booster had, probably accidentally, been turned
on by a plumber, but it was not until a “huge bill” (around £1100/year) arrived that
she realised there was a problem. The occupant reported that she tried and managed
to switch the immersion heater off on the control panel with the help of her father
and the manual. This illustrates the complex issues involved in measuring HP per-
formance. The presence of multiple problems—a low thermostat set-point and the
accidental triggering of resistance heating—might have played a part in lowering the
performance.
60.3.3 Low Load Factor and Performance
CS16 was a newly built farm house of 314 m2gross floor area. The EPC showed
high ratings for all aspects of the dwelling and its systems, and an overall EPC rating
of A. But the monitoring data showed an SPF of only 1.7.
In an attempt to understand the reasons why this HP performed poorly, the
researchers first checked whether the design capacity of the HP (12 kW) was ade-
quate to meet the heat demand of the dwelling. The calculation suggested that there
was sufficient capacity in the HP to heat the dwelling down to external tempera-
tures of around 0 °C. Attention then turned to the monitoring data. It was noted that
the envelope of HP electricity consumption dropped abruptly by roughly 3 kW in
late March 2014, coupled with a drop in the system output temperature. A possible
explanation is that the booster heater was turned off at this point.
Inspection of the mean monthly energy flows in CS16 prompted further questions.
The split between DHW and space heat changed by a large amount between the two
years. DHW went up. Space heat went down by about the same amount. Total heat
output stayed roughly the same.
From the analysis of the interview data, there appeared to be a range of issues
affecting performance, driven by occupant changes to their control strategy. Despite
having photovoltaic panels installed under a generous feed-in-tariff (FIT), the occu-
718 L. F. Chiu and R. Lowe
pants did not feel that these payments were sufficient to offset the cost of their energy
bill. They reported that since the HP had been installed, their electricity bill “goes
up hugely during the winter quarter”, to around £720 in the 2015 winter quarter
and approximately one-third of that in the 2015 summer quarter. They estimated
their annual bill was approximately £1900 that year. In order to save money, they
decided to use the HP only to provide background heating and tried out different
ways to heat the house. These included: turning down most of the room thermostats
to 16 °C (exceptions were the hallways and en suite bathroom, which were set at
21 °C); having their electric oven (an Aga) set to its “snooze” setting (110 °C, with
a corresponding continuous electricity consumption of approximately 600 W) to
provide background resistance heating in the living room–kitchen; and using two
wood-burners (one in the snug and one in the lounge) to supplement the heating
every night in winter, and during the day if it was cold. Moreover, the dwelling was
a farmhouse and the occupants’ life as farmers might also help to explain how lower
temperatures were tolerated as a result of wearing warm clothing inside the house. It
was observed that they and their two dogs spent most of their days in-and-out of the
house. Despite sophisticated zoning, their external doors were opened and shut quite
often, making precise control of temperature in the house difficult. This, coupled
with a rather high electricity consumption resulting from frequent use of appliances
such as a tumble drier, a dishwasher, and a washing machine as well as electric
power showers, makes it unsurprising that their electricity bill was high. But, in the
light of a growing awareness of the consequences of their lifestyle, the occupants
had given up trying to control their internal temperature using room thermostats and
were instead “experimenting” with using secondary heating devices such as the two
wood-burners and the oven mentioned above. It is likely that the use of the Aga,
the power showers, tumble dryer, and the wood-burning stoves, none of which were
controlled by a room thermostat, would have restricted the operation of the HP during
the monitoring period. All of this begins to explain why the HP delivered less than
10,000 kWh/year of heat, less than 40% of the more than 26,000 kWh/year estimated
by the EPC assessor for this house.
Further light was shed on the poor performance of CS16 by a comparison with
two other cases with similar physical characteristics. All three dwellings were over
290 m2floor area (roughly 3.5 times UK mean floor area per dwelling, and the
largest among the 16 case studies). The estimated heat annual demands recorded on
the EPCs for these houses were: CS14, 16,800 kWh/a; CS16, 26,300 kWh/a; CS18,
20,400 kWh/a.
The GSHP installed in CS14 was rated at 12 kW; CS16 and CS18 were fitted
with GSHPs from the same manufacturer. All systems had an integrated auxiliary
heat unit, to back up and supplement the output from the HP. All three dwellings had
underfloor heating. Based on the monitoring data, the mean heat demands of CS14
and CS18 were shown to range between 3.3 and 7 kW in the period from November
to February, yet CS16’s mean heat demand was less than half this, between 1.3 and
2.7 kW. Setting aside the possibility of issues with data logging for the moment, the
60 Ecology of Heat Pump Performance: A Socio-technical Analysis 719
question is whether the low heat demand of CS16 might have contributed to the low
SPF. This working hypothesis was put to the test by plotting the monthly mean COP
versus load factor for these three cases (Fig. 60.3). The graph suggests that:
•The low SPF of 1.7 for CS16 is in fact likely to be correct and can be explained
by the low heat load factor in this dwelling: the HP performance profile plotted
in this way is more or less indistinguishable from the installation in CS14, which
has an SPF of 3.5;
•Monthly COP rises monotonically with heat load factor up to 0.3 for CS14 and
0.4 for CS18—at higher load factors, monthly COP falls, perhaps suggesting the
onset of electric resistance heating.
Note that the CS16 GSHP had a fixed speed compressor, but it is not known
whether ground loop and primary circuit circulation pumps were fixed or variable
speed. If fixed speed, this would also tend to reduce part load performance.
Thus, it appears that efficiency is a function of load factor, and that it may therefore
be affected by behavioural and social factors, such as a requirement to save energy
or reduce high fuel bills, or the use of secondary heating not metered by the HP
monitoring system.
Fig. 60.3 COP versus heat load factor (monthly means) CS14, CS16, and CS18
720 L. F. Chiu and R. Lowe
60.4 Discussion
Using a socio-technical approach that aims to go beyond simply understanding either
users’ behaviours or the technical system, this paper attempts to illustrate the com-
plex ecology of interactions and relationships within and between both systems
in determining performance. The metric of HP performance is itself the product
of a socio-economic-technical system (the Renewable Energy Directive) that sets
the boundary between good and poor performance. Each HP’s operational perfor-
mance is determined from data collected by a monitoring system, the performance of
which depends on quality of installation. The detailed configuration/installation (i.e.
arrangements of physical components) of the HP in the dwelling prefigured occu-
pants’ heating/ventilation practices. These determine the heat load, which impacts
on the HP performance. This in turn appeared to reinforce the original behaviour
of the occupants. The plot of performance against heat load factor provided a way
to encapsulate the dynamic relationships between a constellation of contextual fac-
tors, such as thermal efficiency, lifestyle, perceived comfort, occupants’ practical
understanding and skills in controlling the HP through manipulating thermostat, and
flow rate or the use of non-thermostatically controlled secondary heating, as well
as unintentional human decisions that resulted in an increase of electric resistance
heating. It appears that unexpectedly high electricity bills could be a cue for remedial
measures, triggering actions that led to reduced heat load, and beginning a positive
feedback cycle.
If we take seriously what this study has revealed, we will have to approach policy
implementation, monitoring, and evaluation of low carbon heat technologies differ-
ently. While SPF might be a useful metric for determining the overall success of
policies to support HPs, the complex paths of learning by which performance is
achieved and sustained cannot be ignored. It is possible that an overly simple con-
ceptualisation of uncertainties around HP performance, reflected in energy models,
might have been one factor contributing to the recent scaling back of UK government
support for the RHI programme. A richer understanding of performance in the real
world that captured the process of inter-adaptation between technology and people
[6] might help to avoid such outcomes in the future, and lead to more innovative
policy and more effective interventions to improve performance. For example, many
of the issues raised in this paper could be addressed by reconceptualising heat as a
service and supporting the development of customer care packages.
It has not been possible in this short paper to expand on the theoretical and
methodological developments that have underpinned the analysis, but references
have been given. However, the socio-technical analysis presented here challenges the
conventional separation of physical and social disciplines in applied energy research.
A research programme that opened opportunities for multi-disciplinary collaboration
would be one way to improve the environment for learning.
Acknowledgements The authors acknowledge support from UK Research and Innovation through
the Centre for Research into Energy Demand Solutions, grant reference EP/R035288/1. The analysis
of data from the RHPP Field Trial was undertaken by RAPID-HPC under contract to BEIS, and
60 Ecology of Heat Pump Performance: A Socio-technical Analysis 721
with support from the Centre for Energy Epidemiology, grant reference EP/K011839/1. Support
for the Case Studies Report was provided by Eleni Oikonomou, Colin Gleeson, Jenny Love, Jez
Wingfield and Phil Biddulph.
References
1. HM Government: The Carbon Plan: delivering our low carbon future. Presented to Parliament
Pursuant to Section 12 and 14 of the Climate Change Act 2008 (2011)
2. DECC: The Future of Heating: Meeting the Challenge (2013). https://www.gov.uk/government/
publications/the-future-of-heating-meeting-the-challenge. Accessed 14 May 2019
3. DIRECTIVE 2009/28/EC of the European Parliament and of the council of 23 April 2009 on
the promotion of the use of energy from renewable sources and amending and subsequently
repealing Directives 2001/77/EC and 2003/30/EC. https://eur-lex.europa.eu/legal-content/EN/
TXT/PDF/?uri=CELEX:32009L0028&from=EN. Accessed 14 May 2019
4. DECC: Domestic renewable heat incentive: the first step towards transforming the way we
heat our homes, July 2013. UK Department of Energy and Climate Change. https://assets.
publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/212089/
Domestic_RHI_policy_statement.pdf (2013). Accessed 19 June 2018
5. Renewable Heat Incentive Scheme Regulations 2018. https://www.legislation.gov.uk/ukdsi/
2018/9780111166734/pdfs/ukdsi_9780111166734_en.pdf. Accessed 14 May 19
6. Chiu, L.F., Lowe, R., Raslan, R., Altamirano-Medina, H., Wingfield, J.: A socio-technical
approach to post-occupancy evaluation: interactive adaptability in domestic retrofit. Build. Res.
Inf. 42(5), 574–590 (2014)
7. Lowe, R., Chiu, L.F., Oreszczyn, T.: Socio-technical case study method in building performance
evaluation. Build. Res. Inf. 46(5), 469–484 (2018)
8. Lowe, R., Chiu, L.F., Oikonomou, E., Gleeson, C., et al.: Analysis of data from heat pumps
installed via the renewable heat premium payment (RHPP) scheme. Case Studies Report from
the RHPP Heat Pump Monitoring Campaign, Mar 2017. https://www.gov.uk/government/
publications/detailed-analysis-of-data-from-heat-pumps-installed-via-the-renewable-heat-
premium-payment-scheme-rhpp
