English

This paper describes four simulation models, which reproduce energy and temperature measurements of occupied buildings very well. These buildings represent small to medium size residential low-energy buildings of different construction type, which are typical for mid-Europe. For the simulation, the software TRNSYS is used. The reproduction of measured energy and temperatures in time steps of 15 minutes is well suited for precise predictions on heating energy demand and comfort of the buildings and modifications of them.

This paper describes a new software application for analysing climate data. It is argued that the produced graphical charts can be useful concept stage design tools in their own right. In describing the software this paper presents a review of climate analysis techniques. It is hoped that this review will be of use to the architectural science student. By way of demonstration climatic charts for Limburg, which is close to the conference venue of Eindhoven, are presented. Finally, scope for further development of the software tool is outlined.

In 1998, MoD commissioned ANSYS CFX to develop a numerical procedure, based on Computational Fluid Dynamics (CFD), aimed at providing a practical tool which can be integrated into design processes in order to optimise lowpressure fine water spray fire suppression systems. The work, which employs the CFX software, included a sensitivity study which investigated the influence of a number of parameters and compared the predictions with an experiment conducted by the Loss Prevention Council (LPC). This paper outlines parts of that work. Preliminary results were encouraging.

When making design decisions, which can impact on building capital cost and performance, the design team has to be confident that the simulation tool is generating reliable output. Confidence can be gained via several mechanisms. In all cases data quality is of prime importance. Recognising the quality of data, or the lack of quality, and how this impacts on the predictions of the simulation tool is crucial to making informed design decisions.

Certain energy-related building standards make use of simple numeric indicators to describe a building's geometric compactness. Typically, such indicators make use of the relation between the volume of a built form and its surface area. The indicators are then used along with information on the thermal transmittance of the building enclosure elements to evaluate the degree to which a building design meets the relevant thermal insulation criteria.

A simple approach for calculation of the electrical yield from building integrated PV (photovoltaic) systems has been implemented in BSim2002 - a program package for thermal simulation of energy and indoor climate conditions in buildings. The module – SimPv - takes advantage of the building model geometry and solar distribution routines of BSim to simulate solar irradiation on a face fitted with PV. The routines also handle shades from local and distant obstacles. The electric yield is calculated in SimPv from a very limited number of data on the PV system.

This paper describes a relatively simple software tool that has been developed to perform annual hour-by hour energy simulation of a building. It extends earlier work performed for one-zone buildings (Bernier and Randriamiarinjatovo, 2001) to buildings that can be represented by five thermally similar zones. The simulation program has been built using a general equation solver (Klein and Alvarado, 2002). As implied by the title of this paper, the solver is easy to use both for the software developer and the user.

This paper deals with neural networks modelling of HVAC systems. In order to increase the neural networks performances, a method based on sensitivity analysis is applied. The same technique is also used to compute the relevance of each input. To avoid the prediction errors in dry coil conditions, a metamodel for each capacity is derived from the neural networks.