W8UP9 – Calibration of building performance simulation tools for stock modelling and health impact assessment

W8UP9 – Calibration of building performance simulation tools for stock modelling and health impact assessment
29th May 2018 Giorgos Petrou


With the external environment being a key driver of indoor conditions, the unprecedented increase in ambient temperature – a direct result of anthropogenic climate change – raises major concerns around the issue of indoor overheating risk. To ensure healthy and safe homes, whilst minimising the need for mechanical cooling, accurate prediction of indoor overheating risk is required. The Chartered Institution of Building Services Engineers (CIBSE), with the release of Technical Memorandum 59 (TM59), aimed at providing a coherent method of assessing overheating risk for new builds through the use of Building Performance Simulation (BPS) tools, whose dynamic thermal capabilities allow for the prediction of indoor temperatures in any building design. However, the uncertainties relating to the approximations of physical laws that each tool makes along with uncertainties of the parameter inputs suggested within TM59 could hinder CIBSE’s efforts of providing a consistent guidance for predicting indoor overheating even for the worst-case scenarios. The preliminary work of this doctoral research revealed that BPS tool and heat transfer algorithm choice can have a significant impact on the predicted indoor overheating risk. For the rest of this doctoral work, the focus will be on the quantification of the uncertainty relating to occupant-behaviour variability.


  1. Identify the influence of dwelling and occupant characteristics on the summer indoor temperatures using the monitored data from the Energy Follow-Up Survey.
  2. Cluster the dwellings into homogeneous groups based on the indoor temperature profiles and physical characteristics.
  3. For each cluster, create a Gaussian Process meta-model based on the archetype-based stock models developed by Symonds et al. (2016).
  4. Calibrate the input variables against the monitored temperatures based on theory of Bayesian inference for inverse problems.
  5. Predict the indoor temperatures of the English housing stock under future climate projections.

Expected Results

This work commenced at end of 2017. By the end of 2020, the following key results are expected:

  1. An empirically-based estimate of the distribution of occupant-related inputs that could be used for future modelling studies.
  2. A distribution of indoor temperatures under different weather scenarios.