This study presents a theoretical method that can quickly and accurately identify the locations and strengths of multiple constant contaminant sources indoors by using a single or a limited number of ideal sensors. The method was numerically demonstrated and validated by case studies of sixteen scenarios of contaminant releases in a three-dimensional office. The effects of the number and positions of sensors used, total sampling time, and sampling intervals on the performance of identification were thoroughly studied. This study can help to develop methods for identifying multiple sources by using real sensors as well as optimizing the layout of sensors.

This study presents a theoretical method that can quickly and accurately identify the locations and strengths of multiple constant contaminant sources indoors by using a single or a limited number of ideal sensors. The method was numerically demonstrated and validated by case studies of sixteen scenarios of contaminant releases in a three-dimensional office. The effects of the number and positions of sensors used, total sampling time, and sampling intervals on the performance of identification were thoroughly studied. This study can help to develop methods for identifying multiple sources by using real sensors as well as optimizing the layout of sensors. INTRODUCTION In case of accidental or intentional releases of hazardous contaminant indoors, such as the biochemical terrorist attacks, epidemic outbreak, and toxic gas leakage, quickly identifying the characteristics (e.g., location and emission rate) of contaminant source in short time is critical for taking prompt response measures to protect occupants and mitigate losses. The identification of contaminant source is an inverse problem compared to the prediction of contaminant dispersion,. Although much work has been conducted on the inverse problems in heat transfer (Alifanov, 1994), groundwater transport (Mahar and Datta, 2000), and atmospheric constituent transport (Seibert and Frank et al., 2002), only a little work has been published on the determination of indoor contaminant source. Liu and Zhai (2007) thoroughly reviewed various pollutant inverse modelling methods for both groundwater and air fields. The review indicated that although contaminants in groundwater and air follow the same transport rules, there are still great challenges related to air applications due to the significant property disparities between the two problems. In practice, there are various ways for contaminants to be released into indoor environment. First, the number of sources may be single or multiple. Second, the releases may be instantaneous, or continuous with constant/changing rate. In addition, in some cases, the potential locations of sources are know, while in other cases the potential locations of sources may be totally unknown. For example, in terrorist attacks, hazardous agents may be released at any indoor locations. As a variety of scenarios exist, the research on source identification in indoor environment is full of challenges. In recent years, several studies have been devoted to identify contaminant sources indoors. Sohn et al. (2002) used Bayesian probability model to identify the contaminant source in a five-room building. Arvelo et al. (2002) employed the genetic algorithm to locate the sources in a building with nine offices and a hallway. Zhang and Chen (2007a) used an inverse computational fluid dynamics (CFD) model with quasi-reversibility (QR) equation to identify contaminant source in an aircraft cabin and an office. They further solved inverse contaminant transport model with pseudo-reversibility (PR) method, and compared the PR method with QR method (Zhang and Chen, 2007b). Liu and Zhai (2008) proposed a probability-based CFD modelling method for identifying the location of an instantaneous source. They further developed a probability-based inverse multi-zone modelling method for identifying source location in buildings with many compartments (Liu and Zhai, 2009). The above studies have laid a solid foundation for in depth research of more complex and realistic indoor source identification tasks. However, in these studies, only very few attempts have been made to the problems related to multiple sources. This study aims to develop a theoretical method for quickly identifying the locations and strengths of multiple constant contaminant sources by limited number of ideal sensors. With case studies of 16 scenarios of releases in a three-dimensional office, the performance of method is tested by using different layouts of sensors, total sampling periods, and sampling time intervals. SOURCE IDENTIFICATION METHOD Overview of the method The problem under study is specified with the following assumptions: 1. The indoor airflow field is steady and the contaminant can be treated as passive gas. For most ventilated indoor environments, the airflow field can reach steady-state much faster than the Proceedings of Building Simulation 2011: 12th Conference of International Building Performance Simulation Association, Sydney, 14-16 November.

In the preceding companion paper (Part I), a method with one sensor that could identify the indoor contaminant source location and strength in short time was presented. On the basis of further theoretical study, a method with two sensors is presented in this paper to identify contaminant source with higher accuracy. This paper demonstrates how to use the method with two sensors to find the location of contaminant source in a threedimensional room. In addition, the accuracy of two types of methods was compared. The correctness probability, which are used to evaluate the accuracy of source locating, of the method with two sensors and the method with one sensor are 83.3% and 94.8% respectively. The results show that the method with two sensors works better for locating contaminant source than that with one sensor. In practice, the method with two sensors may be more applicable for the situations where the accuracy of source identification is crucial while the costs of sensors are not of great concern.

From the status quo and future of chemical, biological and radiological (CBR) terrorism, it is pointed out that airborne CBR attacks have posed severe threats to underground buildings. The new problems brought to underground engineering environment security by CBR attacks are Summarized, the possible problems of underground engineering in China are analysed in details, it is pointed out that current defensive ideology and defensive ventilation measures of underground buildings in China don't meet the requirement of defensive CBR attacks. The characteristics of CBR terrorist incidents are generalized, furthermore, the problems of underground engineering environment posed by CBR attacks from the aspect of prevention, treatment and post event remedy are discussed.

Considering the current state of HVAC web sites in the world, this paper points out the necessity of applying WEB database based ASP technology, presenting the main features, methods and advantages of this technology. According to the practical experience, it also provides the basic method to construct a web site of HVAC and specialized information database. Finally, it proposes the prospects of applying internet in the HVAC field.