Energy Efficiency Standard for Social Housing: peer review

5 Additional Retrofit Examples

As noted in section 3 of this report and in the Peer Review brief, additional retrofit examples were required in order to provide a more thorough cross-section of social housing property types. Twelve additional retrofit examples were prepared; these are summarised in Appendix H. The selection process took account both of property types not included in the EESSH consultation document, and of the need to represent hard-to-treat (or expensive-to-treat) properties. The final list of property types was agreed after negotiation with the Scottish Government.

5.1 Additional retrofit examples

The table below details the twelve additional retrofit examples, specifying properties that have been double-tested using both default and calculated specifications.

Table 4: Additional Retrofit Examples

	Hard-to-Treat ( HTT) Property type	Sub-archetype	Default or calculated U-values applied
HTT1	Pre-1919 tenement flat	Ground floor, gable end	Both
HTT 2	Pre-1919 tenement flat	Top floor, gable end, partial flat roof	Default
HTT 3	Multi-storey flat	Ground floor, gable end	Default
HTT 4	Multi-storey flat	Mid-floor, gable end	Default
HTT 5	External deck access flat	Ground floor, gable end	Default
HTT 6	External deck access flat	Mid-floor, gable end	Default
HTT 7	No Fines concrete flat	Ground floor, gable end	Both
HTT 8	No Fines concrete flat	Top floor, gable end	Both
HTT 9	Swedish timber house	Semi-detached	Both
HTT 10	1950-1982 timber house	Semi-detached	Default
HTT 11	System-built concrete house	Semi-detached	Default
HTT 12	BISF steel frame house	Semi-detached	Default

For the flats, two sub-archetypes were selected, in order to provide a level of comparison without the excessive detail of modelling all possible sub-archetypes. In most cases, flat sub-archetypes ( e.g. floor level, gable end) were chosen to illustrate the worst-performing sub-archetypes. However, this is not the case for mid-floor flats: while top-floor flats would achieve a worse baseline, mid-floor flats were generally selected in order to be consistent with the existing retrofit examples. Mid-floor flats are also more prevalent.

Two timber house archetypes were chosen, as the Swedish timber house type generally achieves a higher baseline energy efficiency rating than standard older timber frame houses.

As with the original retrofit examples, each of the above property types was modelled with both gas and electric heating.

All twelve new retrofit examples were modelled using the new version of RdSAP (9.91). Due to the constraints of the software, many of the different non-traditional house types are likely to be treated similarly by RdSAP. In order to provide further analysis and validation in this respect, a third of the examples were modelled twice, once using the 'basic' version of RdSAP and once inputting the optional additional data elements ( e.g. actual U-values, insulation specification, etc.), in order to identify any key discrepancies.

5.1.1 Commentary on additional retrofit examples

A full list of retrofit measures, applied across different standard periods is provided in Appendix H. An overview of measures and outcomes for the full suite of the original 23 retrofit examples as remodelled and the additional example dwellings from this chapter is provided in Appendix H. Further commentary is provided below.

5.1.2 Non-traditional construction

Non-traditional construction accounts for a large proportion of the additional retrofit examples. While it was felt important to cover these property types, it should be noted that the examples are by necessity generic and in most cases were modelled on a worst-case scenario basis. In reality, many landlords may find some or all of their non-traditional stock has been improved since it was built, so their starting baseline may differ from that shown in the examples. (However, it should also be noted that many such improvements may have been more for aesthetic than energy efficiency reasons, e.g. brick overcladding to improve the appearance of the façades.)

Also with regard to non-traditional stock, the limitations of RdSAP must be recognised. As mentioned above, some of the different non-traditional construction types may be treated similarly by the software and so the variations could be limited. Different concrete construction types in particular may suffer from this; for example, system-built properties which could encompass a wide range of stock types would effectively be treated the same by RdSAP.

5.1.3 Flats

With regard to the flats covered by the additional retrofit examples, two sub-archetypes were selected in each case for reasons outlined above. Clearly, archetypes not included will perform differently and realise different savings from refurbishment. The inclusion of partial flat roofs on the pre-1919 flats was felt to be important, as this is particularly common in the later ( e.g. Victorian) traditional tenements. In these properties the roofs are commonly either mansard or a mix of pitched and flat, in many cases the flat-roof area accounting for around two thirds of the total roof area. This is often overlooked when considering traditional tenement improvements, but it is important to bear in mind as flat-roof insulation is notoriously complex and rarely undertaken unless the roofs are being renewed.

5.2 RdSAP Remodelling with non-default values

As noted above, some of the retrofit examples were modelled twice, both excluding and including the additional data elements offered by RdSAP. It is important to reiterate that these additional elements are optional only, and as their inclusion will require additional time by the assessor and possibly additional costs, it seems likely that in most cases they will not be included in the survey.

As detailed in the table 4 in ( Section 5.1 above), the following retrofit examples were modelled twice, using both default and calculated U-values and insulation specifications:

• Pre-1919 tenement flat ( HTT1)- ground floor, gable end
• No Fines concrete flat ( HTT7)- ground floor, gable end
• No Fines concrete flat ( HTT8)- top floor, gable end
• Swedish timber frame house ( HTT9)- semi-detached

For each of these comparisons, using specified rather than default values changes the performance ratings and predicted impacts of improvement measures.

5.2.1 Enhance Baseline Example

This can work both ways: for example, the ground-floor No Fines flat achieves a better baseline (1990) performance when using actual specified values, however this improved starting point then lessens the impact of the improvement measures, so the benefits of the improvement measures are proportionally greater when using the default values (even though these predicted benefits are unlikely to be achieved in reality.)

Again, the pre-1919 tenement flat shows noteworthy discrepancies when using default or actual specifications and capital costs. Examples include the following:

• Concrete floor - RdSAP's default U-value for an uninsulated concrete floor is 0.5, however in situ measurements have recorded U-value as poor as 3.5 - a considerable variation. Indeed, this variation is so large that RdSAP is unable to input the actual U-value: its range is from 0.1 - 2.3, so 2.3 was used as the calculated U-value. Post-improvement, the default U-value is 0.28 (for an unspecified insulation material or depth), in comparison some in situ measurements have shown a U-value of 0.6 - again, a particularly significant difference given that this measured improved U-value is shown to be worse than the default unimproved U-value.
• Windows - For single glazing, the in situ measured U-value is 5.5, compared with RdSAP's default U-value of 4.8. However, the poorest U-value permitted by RdSAP is 5.1 so this was used as the calculated U-value. In situ measurements showed secondary glazing to give a U-value of 2.3; again this differs from the defaults available within RdSAP.
• Walls - RdSAP's default U-value is 2.0, compared with the in situ measured U-value of 1.4.

The tables below illustrate the differences in predicted impacts for an electric-heated pre-1919 tenement flat when using default or calculated U-values. Band changes are highlighted in bold:

Table 5: SAP 9.91 default calculation for pre 1919 tenement

1. DEFAULT	Baseline (1990)	SHQS	2050
SAP rating	44	58	72
SAP band	E	D	C
EI rating	25	34	52
EI band	F	F	E

Table 6: SAP 9.91 user defined calculation for pre 1919 tenement

2. CALCULATED	Baseline (1990)	SHQS	2050
SAP rating	23	39	67
SAP band	F	E	D
EI rating	8	18	45
EI band	G	G	E

For this property, using actual specifications shows the property to be less efficient than would be assumed by RdSAP if default U-values were used, both pre- and post-improvement. Similarly, using actual costs for the secondary glazing and floor insulation (both high-specification products with high associated costs, selected to cater to conservation-grade properties) significantly affects the payback periods. Irrespective of these costs, the choice of whether to use default or calculate specifications is shown to change both the EE and EI bands for this property type - with fundamental implications for whether they meet the EESSH.

5.2.2 Reduced Baseline Example

However, the situation changes in other cases. For the Swedish timber frame house, using the calculated U-values and insulation specifications shows the property to have a better performance both at baseline and post-improvement than would be predicted by relying on the default figures. Some key points in relation to this property:

• The default wall U-values (baseline 2.0, improved 0.6) were based on an assumed baseline performance and application of either an unknown internal or external wall insulation system. This is in comparison with the calculated U-values (baseline 1.24, improved 0.25) which were based on record drawings of the wall construction and specification of insulation materials.
• Similarly, the default improved floor U-value (0.35) was based on application of an unknown floor insulation retrofit, while the calculated improved U-value (0.17) was based on specified 170mm insulation.

The inclusion of specified values is therefore significant. In some cases the differential between the default and specified values actually pushes the property into a different SAP or EI band. For example, the following tables show predicted performance for the (electric-heated) Swedish timber frame house property using 1) default values and 2) specified values. Band changes are highlighted in bold:

Table 7: SAP 9.91 Default calculation for Swedish timber frame house

1. DEFAULT	Baseline (1990)	SHQS	2020	2050
SAP rating	22	52	66	73
SAP band	F	E	D	C
EI rating	7	28	44	51
EI band	G	F	E	E

Table 8: SAP 9.91 User defined calculation for Swedish timber frame house

2. CALCULATED	Baseline (1990)	SHQS	2020	2050
SAP rating	29	59	70	79
SAP band	F	D	C	C
EI rating	13	36	50	60
EI band	G	F	E	D

Tables 7 and 8 show that using the calculated values improves the SAP rating in three cases ( SHQS, 2020 and 2050). This is particularly important when the property's rating is close to achieving the next SAP band and could be the difference between reaching or missing the standard. Where there is documentary evidence, the landlord's assessor will have the option to specify values that vary from the RdSAP defaults which could mean less effort is required to achieve the standard.

5.2.3 The implications of using non-default U-values

Overall, the consequences of using default U-values are not always clear in terms of meeting the EESSH. In some cases, inputting the optional additional data will show improved performance over the default predictions - and therefore be beneficial to the landlords - but in other cases it could shift the baseline and thereby reduce the impact of the improvement measures -making it harder to meet the standard. It seems likely that in many cases default values will be used for the baseline performance, with calculated values being entered for the improvement elements.

To help make predicted impacts more accurate in the future, the Scottish Government may wish to investigate the possibility of promoting or requiring the use of calculated rather than default U-values for certain building elements of improvement measures, where there is documentary evidence that they are materially different. Such an approach could draw upon the Department of Energy and Climate Change's ( DECC) methodology for Green Deal and ECO of in-use factors (although this applies to all measures rather than just some). For a Green Deal Assessment, 'in-use factors' will have to be applied to savings predicted by RdSAP to make the predictions more realistic. These factors will be reviewed and updated over time as new research and monitoring results become available ^[21] . Subject to certain changes (see next paragraph) it may be possible for the Scottish Government to adopt a similar approach for measuring performance of social housing stock using RdSAP: a set of calculated baseline and post-improvement U-values could be drawn up for different building and improvement elements. Social landlords could use these wherever appropriate in place of the RdSAP default figures. These calculated U-values would be obtained from current research ^[22] , and if necessary additional in situ testing, allowing them to be updated periodically as new research and monitoring results become available. This mechanism would limit the gap between predicted and achieved performance, reinforcing social housing's contribution to Scottish Government targets.

5.2.4 Evidence supporting the use of non-default U-Values

The feasibility of this is largely dependent upon what constitutes 'documentary evidence' of revised U-values. At present such 'documentary evidence' must be either 'relevant building control approval' or a calculation 'produced or verified by a suitably qualified person' ^[23] . This presents a potential barrier. Firstly, it requires documentary evidence for each specific building. Secondly, it is likely to require a 'suitably qualified person': this means being a member of a scheme recognised by an approved domestic assessor protocol organisation or other professional body. At the time of writing it is not clear which assessors trained in Scotland are approved to do this under their scheme protocol. If the requirement for building-specific evidence was modified and robust research reports were to qualify this evidence this would aid assessors and other professionals ( e.g. architects). In the longer term the calculations behind the default U-values within RdSAP are likely to be updated ^[24] , however it is uncertain when these updates will take account of the above U-value research and it may prove beneficial for the Scottish Government to consider work focused on common social housing archetypes.

Subsequent to this modelling, DECC published finalised guidance on the operation of the ECO. This states that only types of solid wall insulation achieving a U-value of 0.3 to be eligible for support under CERO (which focuses on SWI). This is significant as the default 50mm insulation in RdSAP only achieves a U-value of 0.6. When designing and modelling a project, social landlords will have to specify a default of 150 mm of insulation to achieve a U-value of 0.25. This option is available in the new version of RdSAP 2009 v9.91. The Scottish Government and housing associations should be aware of these issues and monitor whether any complications arise when modelling specific products for the purpose of seeking ECO.

5.3 Other observations

The additional retrofit examples do not cover housing stock that is protected by planning regulations, e.g. listed buildings or those in conservation areas (as referenced in section 3.2.13). In such cases, the retrofit examples will be of more limited applicability as several of the potential improvement measures may either not be permitted or may require use of specialist systems incurring significantly greater costs. Examples include the following:

• Double glazing - may not be permitted, or may require specialist systems
• External wall insulation - unlikely to be permitted
• Internal wall insulation - may require a specialist system
• Gas boilers - may not be permitted in situations where flue would be sited on principal elevation

Guidance on costs and application of more specialist measures for such properties are available from other sources ^[25] .