Building Performance Prediction: Making Room for the Gap?


Within European Union member states, building energy performance is addressed by the EU Energy Performance of Buildings Directive. One requirement of this document is that all Member States adopt a methodology to calculate building energy performance, which as a minimum accounts for all energy used within a building for the purposes of “heating, hot water heating, cooling, ventilation and lighting”.  Such methodologies act as the foundation for national building energy performance certification schemes, also mandatory under the EPBD, which are required to produce an energy performance certificate whenever a building is constructed, sold or rented.

The majority of member states have opted for an approach of whole-building modelling, often using thermal simulation software. Within the UK, the designated calculation methodology, and subsequent generation of energy performance certificates is conducted using the Standard Assessment Procedure (SAP), and associated software. Such energy assessments typically focus heavily on theoretical performance, and have come under heavy criticism over the past few years in both industrial and academic literature as a key component of the much documented ‘performance gap’ or the difference between predicted and measured energy performance within a building. Virtual building models, the literature argues, are too idealised, with simplifications currently going unrecognised within the design process, and inputs errors going unrecognised.

The ACCEPT team are currently engaged in undertaking a literature review of published material on the performance gap, which represents a core area of research within the project, researching common contributing factors and causes. We thought we’d share some of our initial findings with you ahead of time.

It is often impossible to know upfront exactly how much energy a building will use in operation, due to numerous implicit uncertainties during preliminary design stages. Consequently, modelling software must make a series of assumptions during calculations, with respect to factors such as plug loads, climate conditions, internal heat gains, and user behaviours and schedules. However, performance gap literature reviewed within our study consistently highlights how these assumptions may represent a significant contributing factor to the performance gap; often being highly simplistic, and poorly representative of real world conditions. As such, the literature believes these assumptions are ultimately acting as inaccuracies within the calculation methodologies, which are rarely challenged and which act to produce inaccurate building energy performance predictions; failing to account for asymmetric energy use between different builds and from different building users, typical hours of use, and installed density of appliances.

Throughout the modelling process, it is widely accepted that energy performance assessors are very much reliant on input data obtained through third parties. Whilst there is an emphasis on traceability within the majority of these methodologies, there is growing concern which brings into question the accuracy of this input data, in particular regarding building material thermal performance and U-values. The most common causes of errors in input data stem from the overestimation of published performance values by manufacturers, whether for marketing purposes or as a result of overoptimistic material testing results. Material U-values are often obtained through laboratory tests or numerical simulations. Such assessments are often undertaken on isolated elements, with the obtained figure for ‘technical potential’ expected to work under replication in real world scenarios. However, much like the assumptions within modelling calculations, these figures are often generally optimistic when introduced to the nuances of the real world, and where individual elements must interact to produce one effective building envelope.

Outputs from building energy use calculations are therefore dependent on a competent analyst, accurate and representative third party data, and the appropriateness of the model to the particular context. In addition to these issues, our research has also brought to light a consensus within the literature regarding issues such as;

  • Poor drawings or specification provided by the design team to the energy assessor.
  • Heating systems and associated controls installed differently to that communicated within the design stage calculation.
  • Specified junction details and qualities not adhered to throughout construction.
  • Alterations to the initial building design and/or product substitution not communicated to assessor throughout construction.
  • Windows installed throughout construction are not reflective of the BFRC figures provided to the building modeler.
  • Secondary heating installed without this being communicated within the modelling stage.

With this in mind, it is hardly surprising that predictions can generate a gap downstream in the construction process where energy usage is measured by meters and monitors; where real world pressures kick in, and where assumptions and model inputs are put to the test. The published literature has published case studies across Europe indicating that calculations produced using EPBD compliant software are producing discrepancies of up to 30% between predicted and measured energy performance. Whilst building energy use calculations are not the only area of construction where factors can contribute to the performance gap, it will be important to assess such issues as they are identified if legislation is to be successful in reducing the contribution of the built environment to global emissions and energy usage.

Credit: Zero Carbon Hub 2016

2 thoughts on “Building Performance Prediction: Making Room for the Gap?

Comments are closed.