Assessing the number of users who are excluded by domestic heating controls

Abstract

Space heating accounts for almost 60% of the energy delivered to housing which in turn accounts for nearly 27% of the total UK's carbon emissions. This study was conducted to investigate the influence of heating control design on the degree of ‘user exclusion’. This was calculated using the Design Exclusion Calculator, developed by the Engineering Design Centre at the University of Cambridge. To elucidate the capability requirements of the system, a detailed hierarchical task analysis was produced, due to the complexity of the overall task. The Exclusion Calculation found that the current design placed excessive demands upon the capabilities of at least 9.5% of the UK population over 16 years old, particularly in terms of ‘vision’, ‘thinking’ and ‘dexterity’ requirements. This increased to 20.7% for users over 60 years old. The method does not account for the level of numeracy and literacy and so the true exclusion may be higher. Usability testing was conducted to help validate the results which indicated that 66% of users at a low-carbon housing development could not programme their controls as desired. Therefore, more detailed analysis of the cognitive demands placed upon the users is required to understand where problems within the programming process occur. Further research focusing on this cognitive interaction will work towards a solution that may allow users to behave easily in a more sustainable manner.

Keywords:

• controls
• heating
• design exclusion
• inclusive design
• hierarchical task analysis

1. Introduction

UK housing is responsible for 27% of the UK's carbon emissions (Sustainable Development Commission 2006, Boardman 2007, Lomas et al. 2009) and of this figure, space heating accounts for 60% (Utley and Shorrock 2008). The energy consumed in maintaining indoor temperature depends on four physical factors: the efficiency of the heating system, the efficiency of the building fabric, the temperature difference between inside and outside, and the duration of heating input (Lomas et al. 2009, MacKay 2009). Designing a building in a sustainable manner, however, does not guarantee that the building will be energy efficient as consumption is heavily influenced by the behaviour of its occupants (Derijcke and Uitzinger 2006).

Thermal comfort is highly subjective and simply having some control over the internal conditions (i.e. being able to open a window) generally gives occupants a greater sense of satisfaction (Bordass and Leaman 2001). Typically, a domestic heating system will include a central digital interface and dials at each radiator. This gives occupants some control over how much heat energy is delivered, where in the house it is delivered and when. If an occupant wishes to reduce their domestic consumption, their ability to do so will in a large part be dictated by the design of their heating control systems. These controls may be unnecessarily complex, therefore excluding users from operating them, even at a basic level.

Some commentators argue that simpler, more usable systems could provide for a double-dividend: greater thermal comfort and reduced energy consumption (Bordass and Leaman 2001). Gupta et al. (2009) agree that when programmed effectively controls can save substantial amounts of energy. One study cited by them estimates that up to 50% of US households use their thermostats as on/off switches and not to control temperature. Miller (in Lomas et al. 2009) concurs that one of the best ways of reducing domestic energy consumption is encouraging proper use of heating controls by users. Simplification of these interfaces may encourage proper usage, in particular by focusing on levels of comfort rather than temperature (Gupta et al. 2009).

Control systems should be designed such that ‘environmentally preferred behaviour is also the most logical and easiest accomplished’ (Derijcke and Uitzinger 2006). But ‘logical’ to whom? Perhaps the biggest design challenge is how to accommodate the wide range of physical, sensory and cognitive abilities of the entire population, as Boardass et al. (2007) state ‘well-designed controls with good user interfaces benefit everyone’.

The aim of this study is to quantify how exclusive the heating controls used at a specific housing development are. This is to gain insight into where design improvements could be made and to compare the calculation results with the actual capabilities of users.

2. The controls assessed in this study

The design under study is the Salus RT500 controller used within the domestic environment to control the heating system. It does not control hot-water consumption within the home. Both the duration and temperature of the heating can be specified. Once the user has located the control, they are required to open the control panel door and select whether they want to set the time and temperatures for the weekdays or the weekend. This is done using the arrow buttons and the select button, as shown in Figure 1. For each of five time intervals, the temperature needs to be specified, again using the arrow buttons and the select button. This is then repeated for both sets of days. Once a temperature has been specified for each time interval, the set button is pressed to ready the system and the door is closed. Relevant dimensions of the interface are also shown in Figure 1.

Figure 1 Drawing and measurements of the interface.

Importantly, the aim of this study is not to criticise the design of one particular control but to gain an understanding of design exclusion issues in energy management control design. The lessons learnt from this study will inform a systematic study of wider scope with a greater range of programmable heating controls. By identifying these issues, any subsequent design interventions will address these specifically, resulting in a more inclusive and usable solutions.

3. Assessing user capabilities within inclusive design

Inclusive design is defined as making a product (or service) accessible to the widest range of people as possible, and it recognises that one particular design solution will not include or satisfy all users (Keates and Clarkson 2003).

Assessment methods used in inclusive design fall into two categories: those which involve users directly and those that do not. Methods which involve user participation, such as user observation, interviews or focus groups, can prove expensive and time consuming yet are seen to be more realistic (Cardoso et al. 2004). Methods which do not involve users, such as simulation, task analysis and self-observation, are used to gain insight into problems at specific stages of the interaction (Cardoso et al. 2004). The Exclusion Calculator developed by the Engineering Design Centre at the University of Cambridge falls into the second category of assessment methods and can be used to estimate the number of people currently excluded by a product.

There are several assessment methods available to verify the ‘inclusivity’ of a product or system, such as user trials, ethnography and expert appraisal, which can make explicit the types of demands placed upon the user. Capability simulators can help designers understand the reduced capacity to perform a task from a user's perspective in a cost-effective manner. These simulators can recreate a reduction in certain motor and sensory capabilities, nevertheless it is particularly difficult to simulate a loss of cognitive function accurately, if at all (Goodman and Waller 2007). Despite this, Clarkson et al. (2002) argue ‘it would be ideal if the designer understood the capabilities of the target population prior to designing the product’.

The Exclusion Calculator is intended to help inform decision-making at the beginning of the design process and to work in parallel with other tools to ensure a more holistic design approach (Waller et al. 2009b). The main advantage of the tool is the numerical data produced in its results. These data are originally established from the Office of National Statistics 1996/1997 Disability Follow-up Survey capability scales (Grundy et al. 1999, Clarkson et al. 2002, Dong et al. 2002, Cardoso et al. 2004). The Disability Follow-up Survey uses 13 capability categories to assess levels of impairments, seven of which directly relate to product interaction; seeing (vision), hearing, intellectual function, communication, locomotion, reach and stretch, and dexterity (Waller et al. 2009a,b). In the Exclusion Calculator, the intellectual function and communication abilities have been combined under thinking capabilities.

Capability demands have been defined as the level of ability required to achieve the task (Waller et al. 2009a,b). Understanding this level allows the quantification of user exclusion. According to Card et al. (1983, cited in Keates and Clarkson 2003), human product interaction can be described in three phases: perception, cognition and motor functions. ‘Vision’ and ‘hearing’ are classed as sensory capabilities, ‘thinking’ is classified as cognitive capability and ‘dexterity’, ‘locomotion’ and ‘reach and stretch’ as motor capabilities.

The Exclusion Calculator considers how demanding each task is using a Likert scale from low to high demand for each capability (Goodman and Waller 2007). The level of demand required is correlated with the number of people who would find the task impossible according to the data from Grundy et al. (1999) giving an overall percentage of the population excluded. The tool was published in the inclusive design Toolkit and is publicly available at http://www.inclusivedesigntoolkit.com. Other consumer products that have been assessed based on the data from the Disability Follow-up Survey include:

• mobile phones (Waller et al. 2009b);

• kettles (Dong et al. 2002);

• heating controls, in terms of their visual and dexterity requirements (Etchell et al. 2004) and to compare the effectiveness of inclusive design tools (Cardoso 2005), not to suggest design improvements; and

• digital television across three stages of the life cycle; getting started, basic use and advanced use (Klein et al. 2003).

4. Method – Design Exclusion Calculation

There are two steps to be carried out prior to the calculation itself: deconstructing the task into its constituent parts and ascertaining the level of demand required to complete the task.

A hierarchical task analysis (HTA) was conducted to clarify the tasks required to programme the control. HTA has been a central method in ergonomics research for the past four decades and has the significant advantage of evaluating both the cognitive and physical elements of any task (Stanton 2006). HTA works by breaking down a task into its individual parts and identifying which parts of the task may result in errors. The HTA, shown in Figure 2, shows the 27 decision tasks, as well as a range of physical and sensory tasks, which must be completed in a specific order to achieve the goal of programming the control for a whole week.

Figure 2 HTA of the controls.

Although many of the individual tasks were physically similar (e.g. pushing a button), the complexity of the system lay in the cognitive element of the task. The plans on the HTA illustrate the cognitive processes (decision tasks are shown in the diagram in the diamond-shaped boxes), while the rectangular boxes represent tasks of a physical nature. In order to achieve the overall goal of heating the home, it is necessary for the user to complete all of these tasks in order from left to right.

Ascertaining the level of demand required to complete the task is done using the Exclusion Calculator itself. This requires the analyst to choose between generic demands, such as reading text or recognising a person at distance, and then setting the appropriate level of demand. In some cases, the level of demand is difficult to judge, however, it can be set along the scale between two demand examples. For example, the dexterity required to open the control panel door is felt to be between picking up a safety pin and holding a pen. The calculation is based on a subjective analysis of the capability demands of using the control which may cause variable results and induce errors. Experience of the analyst is therefore critically important. Table 1 details the options selected and the justification for the level of demand set by the lead author.

Table 1 Assessment of the capability demands of the control.

Capability	Type of demand	Level of demand	Reason for choice
Vision	Reading text at various distances	Read ordinary newsprint	Small instruction text inside the door and small size of text on the digital interface
Hearing	None	None	The system has no audio feedback
Thinking	Think clearly without muddling thoughts. Do something without forgetting what the task was while in the middle of it. Tell the time of day without any confusion. Count well enough to handle money. Remember a message and pass it on correctly	Not applicable	The thought process primarily has to deal with sequences and number. These phrases were judged most relevant to the scales available
Dexterity	Performing fine-finger manipulation with either the left or right hand	Between ‘pick up a safety pin’ and ‘use a pen’	To open the control panel door, the top and bottom of the door must be gripped and then pulled to open and pushed to close
Reach and stretch	Reaching one arm out for a long period	Reach one arm out in front (for long periods)	Controls are manually operated and situated in front of the user
Locomotion	Walking various distances on level ground	Below ‘Walk 50 m without stopping’	Transfer to control system is likely to be less that 50 m

5. Results of the Exclusion Calculation

According to the Exclusion Calculation results the controls currently exclude approximately 9.5% of the UK population aged between 16 and 102 (see Table 2). This increases dramatically to 20.7% for the sector of the population that is over 60 years old (see Table 3). This is broken down by the type of capability requirement as follows in Tables 2 and 3, with thinking, vision and dexterity being the largest demands placed upon users.

Table 2 Exclusion Calculation results for the UK population aged 16–102 years old.

Capability requirement	Number of people excluded aged 16–102	Percentage of population aged 16–102
Vision	1,525,000	3.4
Hearing	0	0
Thinking	2,070,000	4.5
Dexterity	1,670,000	3.7
Reach and stretch	318,000	0.7
Locomotion	895,000	2
Total exclusion^a	4,327,000	9.5
a Total adjusted by the calculator to account for the overlap between disabilities.

Table 3 Exclusion Calculation results for the UK population aged 60–102 years old.

Capability requirement	Number of people excluded aged 60–102	Percentage of population aged 60–102
Vision	1,009,000	8.6
Hearing	0	0
Thinking	964,000	8.2
Dexterity	981,000	8.4
Reach and stretch	200,000	1.7
Locomotion	580,000	5
Total exclusion^a	2,430,000	20.7
a Total adjusted by the calculator to account for the overlap between disabilities.

The results confirm that a large cognitive demand is placed upon users which became apparent at an early stage through the use of HTA. The advantage of the calculation results are that it allows the most demanding capabilities to be prioritised relative to each other. Furthermore, it is import to consider different age ranges as the prevalence of disability increases with age.

One particular aspect of age-related disability is that it can be ‘slow to develop, but multiple in nature’ (Coleman 2003). In order not to count people with multiple capability loss twice, when the demands are too high for a person to complete tasks, they are marked. Once marked, this person will be excluded from any further results where another demand is beyond their capability. This explains why the total exclusion is not simply the sum of all the excluded people given below in the results table.

6. Discussion of Exclusion Calculation results

The three areas found to be excluding the largest number of people are ‘vision’, ‘dexterity’ and ‘thinking’ requirements. Future design effort should concentrate on trying to reduce the requirements in these areas.

A summary of the most effective improvements includes:

• audio feedback provision;

• larger, more contrasting, buttons;

• a larger, clearly laid out screen;

• improved tactility of the interface;

• simplified programming (including a back or undo button); and

• the removal of the control panel door.

It is important to pay particular attention to the digital interface and the information it conveys. The layout and presentation of this information is crucial in reducing the cognitive demands. Currently, the area of the digital screen accounts for less than 10% of the whole interface which for such a critical part of the display is extremely small. The layout of the information is crowded, the size of the display text is small and there is little visual contrast between the text and the background, all of which place large visual demands upon the user.

From a cognitive perspective, it is not necessarily the number of tasks required that proves difficult but the complexity of the overall task, its repetitive nature and the lack of flexibility within the system. The volume of information provided in such a small space on the digital interface may also increase the cognitive demands on the user, leading to confusion. When a mistake is made there is no facility to go back to a stage, resulting in frustration for the user. The system also requires an understanding of temperature scale and its units of measurement which some users may struggle with due to its somewhat abstract nature.

No audio feedback is provided by the system at present; therefore, there are no hearing requirements. The provision of audible feedback could help users, particularly those with visual impairments. Audio feedback could also be haptic by confirming the current settings of the control which could in turn improve user confidence in the system and encourage adjustment as appropriate.

There are two dexterity requirements to be addressed: the opening of the control panel door and the pressing of the buttons. Opening the control panel door is the more exclusive of the two actions as it requires substantial grip strength from one or both hands, a potentially painful but essential step for the user. Removing the door completely would result in the biggest reduction in exclusion related to dexterity. Even though pushing the buttons does not require a significant level of force and therefore does not have high dexterity demands, improving their contrast and size could reduce visual demands further.

The limitations of the Exclusion Calculation must also be considered at this stage. One major potential source of error is that the calculation is based upon population data from 1997. Although the data may be somewhat dated, no ideal data-set currently exists (Waller et al. 2009a). Undoubtedly, the data will have changed in the ensuing decade, particularly considering the rapidly ageing population in the UK today. Currently, there is a new UK-wide study being conducted by the University of Cambridge, aiming to collect data on disability prevalence specifically relevant to product design (Waller et al. 2009a). The results of the pilot study are expected to be available by the end of 2010 before a full survey is conducted.

Additionally, the calculation results represent the number of people excluded by the product not the number of households. It is likely that someone within the household could potentially use the controls, however, this is not consistent with the social model of disability used in inclusive design. This suggests that a disability is the consequence of society or an environment rather than a physical or mental impairment; therefore, the inability to use a product is a result of poor design and not a medical issue (Imrie and Hall 2001).

7. Relating user exclusion to energy consumption

7.1 Context of study – Elmswell ‘Three Gardens’ housing development

To relate this user exclusion to a real-world context, a study was designed to assess whether or not the occupants at the Elmswell ‘Clay Field’ Housing development could use their controls successfully. The East-Anglian site comprises 13 two-bedroom and 9 three-bedroom houses, plus 4 one-bedroom flats, each constructed to the same design specification and was awarded BRE's EcoHomes Excellent certification. The site systems exceed the requirements of UK Building Regulations and typical UK dwellings. Low in-use carbon emissions and utility consumption is facilitated by the measures highlighted in Figure 3.

Figure 3 Overview of sustainable features of Elmswell.

As part of a comprehensive post-occupancy evaluation study at the site, the heat energy consumption of each occupied dwelling was monitored. The data showed that average annual space heating consumption accounts for 54% of the total energy (space heating, hot water and electricity) consumed within the dwelling. Average space heating consumption was 73 kW h/m²/year. Total heat consumption, including space heating and hot water, within individual dwellings ranged between 46 and 145 kW h/m²/year.

In comparison, MacKay (2009) calculated that a 1940s three-bedroom semi-detached house consumed approximately 185 kW h/m²/year in space heating before an energy efficiency refurbishment and approximately 62.5 kW h/m²/year afterwards. Furthermore, calculations have shown that turning the thermostat down from 20°C to 17°C reduced heating consumption by approximately 30% (MacKay 2009).

7.2 Comparing the Exclusion Calculator results to resident capabilities

According to the Organisation for Economic Co-operation and Development (OECD 2000) report, ‘approximately 20% of the adult UK population have difficulties with basic reading and maths’, implying this alone could exclude around 9 million adults over 16 years old, using 1997 population figures. These people would not perhaps be classed as having a disability and consequently would not be counted under the Disability Follow-up Survey (Grundy et al. 1999). Combining this with the results of the Exclusion Calculation, the exclusion could be in the region of 30% of the UK adult population.

To estimate the true exclusion of the heating controls, the residents were asked to complete a task using the controls while being observed and timed. The participants consisted of 12 adults (11 females and 1 male) who lived in the development. The predominantly female sample reflected the occupants of the houses during the daytime.

During the post-occupancy evaluation, an interview was conducted with residents from 11 of the dwellings. This was divided into general lifestyle questions and then questions regarding occupants' water, heating and electricity consumption. After the section of the interview regarding heating consumption, participants were asked to complete a task using their heating control system. The task was to set their heating controls to match the heating profile given in Table 4.

Table 4 Heating profile occupants were asked to programme.

Day	Time	Temperature (°C)
Monday–Friday	6 am	19
	7 pm	22
Saturday and Sunday	6 am	21
	10 pm	17

Prior to the test, the original control settings were noted and the controls reset to the factory default settings in order to give a consistent starting point. The participants were timed and observed during their attempt of the task. They were allowed to use the product instructions displayed on the inside of the panel door to aid them, however, no assistance was provided by the researcher during the test. Afterwards, the controls were programmed either back to their original settings or to the desired settings of the participant.

Of the 12 participants, 8 could not complete the task; 66.6% of the sample. The average time before participants stopped and gave up was 2 min and 38 s while the four participants who could complete the task did so in an average time of 1 min and 34 s.

Of the participants, four females admitted another member of the household was responsible for the programming of the controls. A further four female participants, who were the sole users of the controls within the house, admitted they did not know how to use the controls before attempting the test. It is pertinent to note that the maximum and minimum consumers on-site both occupied three-bedroom dwellings, which had similar occupancy in terms of the number of occupants and time spent in the house. Furthermore, both participated in the usability testing. Prior to attempting the task, the minimum consumer stated ‘I don't really know how [to use the controls], it's stupid, I just use the up and down buttons’ which, in tangent with a constant 17°C set-point, may help to explain their low consumption. In comparison, the maximum consumer whose set-point was always above 21°C said ‘well I'll tell you now, No I can't use it’ despite being part of the minority who could programme their controls correctly. Additionally, the participant expressed an interest to receive help to programme the settings more efficiently. The intimidating perception of the controls deterred that occupant from making changes to reduce consumption. Furthermore, in the initial lifestyle questions, one other participant stated that ‘I'm struggling to program it [the heating] to come on when I want it to’. This implies that the inability to use their controls was a problem of high priority.

Three common problems participants encountered were

• the controls not being intuitive enough to use without help of instructions;

• participants not entering programming mode and instead resetting the clock repeatedly; and

• pressing the set button instead of the select to attempt to move between time or day settings.

All participants used the instructions as reference and two spent the first 30–40 s of the test reading them before pressing any buttons. The first instruction given in the controls was how to set the system clock and not programming the heating system which came second. This resulted in the second problem of repeatedly resetting the clock rather than entering programming mode. This resulted in two users thinking they had completed the task successfully when they had not.

A further common error made was that participants struggled to move from one stage in the process to the next as they instinctively pressed the set button after they entered the first time and temperature settings. This in fact exited the programming mode and sent the participant back to the start of the process, which commonly resulted in frustration for participants.

Although the sample size was small, it was representative of the occupants of the development, all of whom had all lived there for over 1 year. As a result, the findings are only valid for this development and may not be representative of the general population.

8. Conclusions

The control design under study placed excessive demands on the capabilities of at least 9.5% of the UK adult population, with this exclusion doubling for users over 60 years old according to the Exclusion Calculation. The three most demanding capabilities were found to be vision, thinking and dexterity. Design efforts should centre on reducing these demands as a priority. The calculated exclusion was significantly different from the actual exclusion found at the Elmswell development. Many of the users at this development could not interact effectively with their controls and 66.6% of the sample was unable to complete the programming task.

Furthermore, the literature agrees that efficiently programmed heating controls can save energy, however, usability problems are little understood. More detailed analysis of the cognitive demands placed upon the users is required to understand where problems within the programming process occur. A reduction in the cognitive demands placed upon users ought to make the heating controls easier to use. Whether this improvement in usability will result in a reduction in energy consumption will form the core of future research. By designing a more inclusive control system, heating controls may be used more effectively. As a consequence, consumption could be decreased by reducing both the temperature and duration of heating. With this focus on how people interact with control systems within their homes, a solution that allows users to behave easily in a more sustainable manner may be achieved.