A core drilled through wood fibre insulation.

About this image

Cores were drilled into IWI insulation systems (wood-fibre above) to assess their condition. No signs of deterioration were observed in either the PIR or the wood-fibre insulation systems © Historic England

Simulation Models and Energy Efficiency in Historic Buildings

Are building simulation models a good predictor for understanding the real energy-efficiency performance of traditional buildings?

The assumption of poor energy efficiency performance of traditional buildings is largely based upon the use of predictive modelling tools whose reliability can vary widely.

There is a widely held view that traditional buildings are inherently energy intensive and require significant improvement to reduce their energy use and carbon emissions. The assumption of poor performance is, however, largely based upon the use of predictive modelling tools whose reliability in assessing the real performance of traditional buildings can vary widely depending on the information used for the simulations.

Common simulation software are designed to calculate energy consumption and thermal performance and use hygrothermal assessment, a technique which analyses heat, air and moisture transfer through materials, to evaluate the potential of long-term risk from moisture accumulation after retrofit.

Simulation models - false but useful?

The ability to be able to predict performance is helpful as it can guide policy makers and building professionals in their decision-making.

It is often used as a relatively quick, cost-effective method of understanding the impacts of any interventions on buildings, unlike real-time monitoring which takes time and is resource intensive. However concerns about the usefulness of modelling are well known as all models are abstracts of complex real-world situations and the predictions will be as accurate as the data used when modelling.

The question is, is modelling useful even though it is often a false representation of what is actually happening? And how can we improve its accuracy?

Is it a simple fact that all models are both false and useful, and, if so, how can that be? These are the some of the questions that we have been trying to answer in our research into understanding the actual energy efficiency, thermal performance and hygrothermal behaviour of traditional buildings.

Problems with using standardised data for traditional buildings

The Reduced data Standard Assessment Procedure (RdSAP) is the most well-known but problematic model used to assess the energy performance of traditional buildings as it is the basis of Energy Performance Certificates (EPCs).

EPCs are used by the Government to drive energy-efficiency policy but it is not a model that is based on energy performance: rather, it is a model based on energy costs.

The model frequently underestimates the energy efficiency of traditional buildings and makes recommendations for work that may be neither necessary nor appropriate. The workings of EPCs are substantially based on assumptions, relying on standardisation of data which are not reflective of a building’s actual energy and thermal performance, the context or occupancy behaviour.

This inaccuracy skews any analysis of the overall performance of traditional buildings and it could potentially lead to flawed decisions, to the detriment of the traditional fabric.

Our study into the in-situ U-values (a measurement of heat loss used in EPC’s) of traditionally constructed solid walls has indicated that their actual thermal performance has often been underestimated by as much as a third, and that accurate calculations are very much dependent on the quality of the input data.

If in-situ values are used for EPC’s, care should be taken that the measurements are gathered under suitable conditions, for sufficient duration, and with an understanding of the walls’ construction.

A graph comparing u-values of different types of walling.
A comparison of modelled U-values using a range of default thermal conductivities with 18 measured in-situ U-values, showing that the default values underestimated the real average U-value by a third. © Historic England

Even where good quality data are available, in-situ U-values have their limitations as they give only steady-state mean values and therefore can only be seen as a broad representation of the heat loss from a wall. They do not reflect dynamic real-life conditions. The values will be affected by many factors, from the moisture content of the walls to the climate on the day, the orientation, and the condition of the building. Traditional walls are variable as they are heterogeneous constructions, hence results will vary across one wall.

Understanding heat and moisture in historic buildings

Understanding the hygrothermal performance of traditional buildings is one of the most important steps in creating durable and healthy energy-efficient buildings. The use of hygrothermal simulation tools has increased in response to the growing number of reports of moisture problems after retrofit. We know that to reduce carbon emissions we need more energy-efficient buildings. However, improved energy efficiency may result in problems if inappropriate measures are installed that interfere with the correct management of moisture.

Traditional buildings are particularly vulnerable to unwanted effects of energy retrofit measures.

Their moisture behaviour is completely different to that of a modern construction. Unlike modern buildings, in which impermeable vapour barriers are employed to keep moisture from entering, traditional constructions are composed of hygroscopic and semi-permeable materials and are naturally ventilated, allowing the transfer of moisture vapour to maintain their equilibrium. However, problems can occur when they are altered, particularly by adding modern impermeable materials which produce changes in their hygrothermal behaviour.

The main advantage of modelling is that, if the building fabric has been accurately characterised, the long-term hygrothermal performance of various energy retrofits can be predicted, minimising any potential damage to the fabric or to the occupants’ health. Accordingly it can help to select retrofit strategies, identify risks of interstitial condensation, investigate influence of driving rain and assess the impact of flooding on the environmental conditions inside a building.

Determining the hygrothermal behaviour of traditional constructions and any proposed interventions is complex.

However, determining the hygrothermal behaviour of traditional constructions and any proposed interventions is complex. It requires the evaluation of heat (conduction, convection and radiation), airflows (natural and mechanical) and moisture (vapour diffusion and liquid transport) interactions. It requires detail of the geometry of the building assembly, its materials’ properties and the exterior and interior boundary conditions. Orientation and the degree of shading and exposure are also significant. As these interactions occur dynamically, it can be difficult to predict with any certainty the impact of alterations and its associated technical risks from moisture.

Gaining evidence of the effects of energy retrofit

Unfortunately, very little empirical evidence exists about the long-term effects of energy retrofit on the hygrothermal balance of traditional buildings. Concerns about the risk of moisture accumulation from fabric upgrades to traditional buildings led us, in collaboration with Dr. Paul Baker from Glasgow Caledonian University, to carry out research at three sites:

  • a brick terraced house in New Bolsover, Derbyshire
  • Shrewsbury Flax Mill Maltings, Shropshire
  • and a sandstone building in Appleby, Cumbria

At Bolsover and Shrewsbury, we have been monitoring the moisture conditions at the wall-insulation interface or within the walls after the application of internal wall insulation, and we will be doing the same at Appleby.

Energy efficiency Sensors being installed at a historic building.
Installation of sensors at a building in Chapel Street, Appleby, prior to the installation of lime hemp insulation. © Historic England

The results have been compared with modelling using the 1-D WUFI software - a commonly used hygrothermal modelling tool developed by Hartwig Künzel at the Fraunhofer Institut Bauphysik in Germany - to provide confidence in the gathered results.

Schematic diagram of WUFI model with wood fibre.
Schematic diagram of WUFI model with wood fibre. © Historic England

Further, we have been looking into the factors affecting the accuracy of the predictions by carrying out a sensitivity analysis to assess the relative significance of a range of input parameters and to inform best practice when modelling traditional constructions. From this we were able to analyse the reliability of the model and identify which parameters have the greatest impact on the simulations and create the greatest uncertainties.

 We have been comparing the performance of two types of commonly used internal wall insulation: first, hygroscopic insulation systems such as wood-fibre board sandwiched between a bonding undercoat and a top skim coat of lime plaster; and, second, an impermeable, non-hygroscopic insulation known as PIR or polyisocyanurate (a thermoset plastic rigid board) with aluminium foil on both sides finished with plasterboard.

The former represents an approach which is viewed as more sympathetic for use in traditional solid-walled buildings, benefiting the absorption, transport and release of moisture; whereas the latter represents the more conventional vapour-checked solutions, as used in modern buildings. Where possible, we have been collecting local climate data and data of the inside environment and have carried out laboratory tests to characterise the traditional fabric.

Our observations so far at New Bolsover are:

  • Testing has shown that the thermal upgrades have reduced the thermal performance by around 40%
  • the performance of the two insulation systems are similar with defined seasonal cycles - high relative humidities in the winter and drier in the summer
  •  there are incremental increases in relative humidities year-on-year, which implies that moisture is accumulating at the wall-insulation interface, although we have not observed any signs of condensation or mould nor any reduction in thermal performance. This suggests that, if conditions at the wall-insulation interface are allowed to drop to lower relative humidities during the summer months, the insulation systems will recover sufficiently to manage the higher levels during the winter

A core drilled through wood fibre insulation.
Cores were drilled into IWI insulation systems (wood-fibre above) to assess their condition. No signs of deterioration were observed in either the PIR or the wood-fibre insulation systems © Historic England
  • there are clear influences of the climate (in particular, solar radiation), orientation, exposure and degree of shading. The west elevation which is exposed to the sun is performing better than the south elevation which is relatively sheltered
Graph showing moisture  accumlated at the wall-wood-fibre insulation interface within a historic building.
New Bolsover: measured relative humidities on the first floor at the wall-wood-fibre insulation interface. The graph shows accumulated moisture over the last 7 years, and the effect of the seasons and orientation. The west elevation is best performing as it is the most exposed and receives the greatest amount of sun, whereas the south elevation is sheltered and has high relative humidities.

• when modelled, using the measured results as input data, we found broad agreement for the exposed west elevation but a less good fit for the sheltered south elevation, particularly during the colder months

A graph measuring humidity of the interface between insulation material and walling in a historic building.
Comparison of modelled and measured relative humidites of the exposed west elevation with wood-fibre insulation on the first floor during 2015. The graph shows reasonable agreement between the two datasets, though the simulation overestimates the moisture levels during the summer and underestimates the moisture levels during the first part of the year. © Historic England
A graph showing humidity at the interface between gable wall and insulation material in a historic building.
Comparison of modelled and measured relative humidities of the sheltered south elevation with wood-fibre insulation on the first floor during 2015. The graph shows poor agreement between the two datasets with the exception of the summer period; for the most part, the simulation underestimates the moisture levels. © Historic England

Limitations with hygrothermal modelling

The difference between the simulations and the measured data has highlighted some of the uncertainties with modelling.

Modelling of conditions at both Shrewsbury Flax Mill Maltings and at New Bolsover has demonstrated the importance of accurately estimating the rain adherence fraction, which has an influence in driving rain calculations, and the amount of solar radiation and degree of shading on a wall. All have a significant effect on the accuracy of the simulations.

A graphic showing comparisons of predicted and actual measured humidity in a historic building.
A comparison of the measured and predicted interface relative humidities of the wood-fibre insulation on the south elevation on the first floor during 2015. The graph shows a range of relative humidities that can be achieved by using different rain adherence factors (RA). Generally for the RA values 0-0.075 of the simulation results, whilst tracking the dynamic behaviour of the measured data, they underestimate the latter by about 10% relative humidity. © Historic England

Another limitation of modelling relates to situations where materials are not in direct contact with each other and where there is convective airflow. This was suggested by the noisier data with the PIR insulation on the south elevation at New Bolsover, where there is a service gap between the insulation and the brickwork. Airflow is a three-dimensional phenomenon and the WUFI 1-D is a one-dimensional model.

One of the biggest hurdles in modelling is the lack of historic material data from UK buildings.

One of the biggest hurdles in modelling is the lack of historic material data from UK buildings as the WUFI ‘library’ is comprised of building materials from Germany. We have found that uncertainties of the predictions are greatly reduced when accurate material property data are used.

At Shrewsbury Flax Mill Maltings, we modelled the hygrothermal performance of the south and east elevations using material property data of three different types of bricks: default data from two bricks from the WUFI ‘library’ database, referred to as a historic and a handmade brick, and real data taken from one of the bricks from the Flax Mill. Combining the specific climate effects of solar radiation and rainfall on each wall with the different material property data, we found the library bricks suggested a greater risk of moisture accumulation than the brick taken from the Mill, particularly on the east wall.

A graph comparing simulations of energy efficiency.
A comparison of simulations on the east elevation using real material property data from Shrewsbury Flax Mill Maltings (Ditherington in the graph) and brick data from the WUFI ‘library’ historic and handmade bricks. It shows the effect of material property, orientation and weather on relative humidity of the three different types of brick and reveals that the measured data provides significantly lower moisture levels.

One explanation for this is the different moisture capacity of each brick and its ability to absorb, transfer and release moisture; other reasons are the effect of the climate (the amount of solar radiation and rainfall on each elevation) and the degree of shading.

Graph showing comparison of material testing of different types of brick.
Material testing of the Shrewsbury Flax Mill brick (Ditherington in the graph) has provided lower moisture absorption at high humidities compared to the WUFI ‘library’ historic and handmade bricks. © Historic England

Powerful tools if used with caution

It is advisable to take a cautious approach when modelling as models are full of implicit assumptions. For modelling to be useful, we must question the assumptions underlying the analysis in order to understand the parameters used for the simulations. There are numerous factors that affect the performance of traditional buildings: the interactions are dynamic and are continuously changing. Further, often calculations cannot model defects and are based upon idealised homogenous walls which are in perfect condition. As everyone knows, this is never the case for traditional buildings!

If models are used with care and an understanding of their limitations and are calibrated with real data, they can be powerful tools.

However, if models are used with care and an understanding of their limitations and are calibrated with real data, they can be powerful tools. Though the simulations cannot be viewed with complete confidence, it can help to inform decision-making and provide an indication of risk. As the statistician George Box once remarked, ‘all models are wrong, but sometimes useful’.

Acknowledgements

Thanks are due to:

  • Iain McCaig, Senior Building Conservation Advisor, Building Conservation & Geospatial Survey, Historic England
  • Dr. Paul Baker, Glasgow Caledonian University
  • Bolsover District Council, Derbyshire

About the author

Soki Rhee-Duverne

Soki Rhee-Duverne

Building Conservation Advisor, Historic England

In her current role within Historic England's Building Conservation and Geospatial Survey team, Soki primarily manages research projects on energy efficiency and the hygrothermal performance of historic buildings, particularly in relation to technical risks from moisture. She has published on thermal performance of traditional buildings and their elements, hygrothermal modelling, applications of infrared thermography, use of X-ray fluorescence (a non-destructive analytical technique used to determine the elemental composition of materials) on World War Two collections and environmental management of historic house libraries.

Further reading

Baker, P 2015 Hygrothermal modelling of Shrewsbury Flax Mill Maltings, Historic England Research Report 88/2015

Baker, P 2017 A sensitivity analysis of WUFI simulations.  Interim report for Historic England (publication forthcoming)

Baker, P 2019 Post-intervention monitoring and hygrothermal modelling of end of terrace house in New Bolsover, forthcoming interim report for Historic England

Historic England guidance on Energy Performance Certificates

Huijbregts, Z van Schijndel, A Schellen, H and Blades, N 2014 ‘Hygrothermal modelling of flooding events within historic buildings’, Journal of Building Physics, Issue 38

Karagiozis, A Künzel, H and Holm, A 2001 ’WUFI-ORNL/IBP – A North American Hygrothermal Model’, Proceedings of performance exterior elements of whole buildings VIII, Integration of building elements, December 2-7, Clearwater beach, Florida, USA

Lstiburek, J 2015 ‘Wufi: barking up the wrong tree?’ Building Science Insights 89 

Rhee-Duverne, S and Baker, P 2013 Research into the thermal performance of traditional solid brick walls, Historic England Research Report 70/2013

Rhee-Duverne, S and Baker, P 2015 A retrofit of a Victorian terrace house: a whole-house thermal performance assessment, Historic England Research Report 103/2015

De Senlincourt, K 2015 ‘The risks of retrofit’. Green Building 28, Summer

Sustainable Traditional Building Alliance 2015 ‘A Bristolian’s guide to solid wall insulation’

Sustainable Traditional Building Alliance 2015 ‘Planning for responsible retrofit of traditional buildings’

WUFI website 

Was this page helpful?