Our programme of research delivers a major portfolio of reports and practical guides

Low and zero carbon homes: understanding the performance challenge (NF41)


Evidence for a CO2 performance gap has been accumulating since 2007. The closure of this gap has been given high priority in the run up to 2016 to help ensure that zero carbon homes meet design expectations.

This report reviews the evidence that supports the existence of a CO2 performance gap. It identifies a range of possible causes and issues that may contribute, from the earliest stages of design through to post-construction checking. The significance of these causes is under detailed evaluation in a programme of work being undertaken by the Zero Carbon Hub.

Summary of content


If the energy consumption of an occupied home is greater than its designer predicted, then its carbon dioxide emissions will also be higher than predicted – this is the CO2 performance gap. There appears to be a growing body of research evidence that new housing is failing to deliver the anticipated levels of CO2 emissions, although there is relatively little understanding within the wider industry of what might be causing this.


Studies of housing schemes over the last 30 or more years have provided many useful insights, although they do not agree on the precise causes or the scale of the problem. For example, the design predictions for space heating energy at Milton Keynes Energy Park in the late 1980s showed extremely close agreement with measured consumptions. When low energy housing in Salford was revisited, the average energy consumption was found to be almost exactly the same as had been measured 20 years previously.

More recently, co-heating tests at Stamford Brook, Altrincham, revealed that party walls did not have the zero U-value assumed in the standard design calculations; homes at Elm Tree Mews, York, had a higher heat loss than had been predicted, due to additional structural timber in the manufactured frame; and latterly a co-heating test on the Avante development in Maidstone measured the fabric heat loss as slightly better than predicted.


There are seven key questions that need to be considered in order to understand how a performance gap could arise:

  1. Is the assessment model that was used to make the prediction accurate, and has it been correctly implemented in the software used by the designer?
  2. Is the model’s input data correct (and if not, is that due to the conventions or the user)?
  3. Is the home’s design overly complex, presenting unreasonable challenges to the construction team?
  4. Are there fundamental construction quality and skills issues?
  5. Do building materials and mechanical and electrical (M&E) systems perform as well in practice as laboratory tests predict?
  6. Do changes in specifications get properly communicated?
  7. Are the post-construction tests and checks appropriate and adequate?