THERMAL COMFORT IN RESIDENTIAL BUILDINGS: SENSITIVITY TO BUILDING PARAMETERS AND OCCUPANCY. A.Ioannou 1 , L.C.M. Itard 1 1 OTB, TU Delft, Delft, Holland ABSTRACT Dynamic simulation is widely used for assessing thermal comfort in dwellings. Simulation tools, though, have shortcomings due to false assumptions made during the design phase of buildings, limited information on the building's envelope and installations and misunderstandings over the role of the occupant's behaviour. This paper presents the results of a Monte Carlo sensitivity analysis on the factors that affect the PMV comfort index. The reference building was simulated as both Class-A and F according to the Dutch determination method for the energy performance of residential functions and buildings (ISSO 82.3, 2009), with three different heating systems. The study focuses on the heating period which is of main interest concerning residential energy use in the Netherlands. For the PMV the most influential parameters were found to be metabolic activity and clothing, while the thermostat had secondary impact. INTRODUCTION The international standard ISO 7730 is a commonly used method for predicting the thermal sensation (PMV) and thermal dissatisfaction (PPD) of people exposed to moderate thermal environments. The PMV model predicts the thermal sensation as a function of activity, clothing and the four classical thermal environmental parameters: air temperature, mean radiant temperature, air velocity and humidity. Activity means the intensity of the physical activity of a person and the clothing is the total thermal resistance from the skin to the outer surface of the clothed body. Many widely used building simulation programs such as ESP-r, TRNSYS and Energy+ use ISO 7730 (ISO 2005) to calculate comfort levels inside a building. There is a significant gap in the literature when it comes to sensitivity analysis of physical and occupancy parameters in the residential sector of areas with a cold climate such as North Western Europe. No studies have evaluated these parameters with a complete sensitivity analysis method which reflects the occupant’s behaviour such as ventilation and thermostat settings as well as physical parameters for the PMV comfort index. This paper presents the results of a sensitivity analysis study that was performed for a single residential housing unit in the Netherlands. The simulations were carried out with the following variations: multi-zone and single-zone versions of the building; two different grades of insulation; three different types of HVAC services; the occupant’s behavioural characteristics (thermostat level, ventilation behaviour, metabolic rate, clothing and presence). The sensitivity of the above-mentioned parameters was gauged for the hourly PMV comfort index and the results focus on the heating period. METHODOLOGY The goal of the study is to make recommendations for: 1) Which parameters have the most critical influence on the PMV comfort index? 2) Is the sensitivity different for dwellings with different physical qualities and different energy classes? Sensitivity Analysis The goal of the sensitivity analysis is to study the response of the model simulated by EnergyPlus with respect to the variations of specific design parameters. It can be used to assess which set of parameters has the greatest influence on the building performance variance, and at what percentage. Sensitivity analyses can be grouped into three classes: screening methods, local sensitivity methods and global sensitivity methods. Screening methods are used for complex, computationally intensive situations with a large number of parameters, such as in sustainable building design. This method can identify and rank in qualitative terms the design parameters that are responsible for the majority of the output variability e.g. energy performance. These methods are called OAT methods (one-parameter-at- a-time) and the impact of changing the values of each parameter is evaluated in turn (partial analysis). A performance estimation using standard values is used as control. For each design parameter, two extreme values are selected on either side of the standard value. The differences between the results obtained by using the standard value and the extreme values are compared in order to evaluate which parameters Fifth German-Austrian IBPSA Conference RWTH Aachen University - 75 -