Research Projects

Real-time Locator of Contaminant Sources (LOCS) in Indoor Building Environments

A growing interest in building security and occupant exposure to contaminants revealed a need for fast determination of indoor pollutant source locations during incidental situations, such as terrorist attacks or accidental contaminant releases. Current state-of-the-art methods include: (1) genetic algorithm based classifications, (2) matching of accidental contaminant distributions with cases from pre-computed databases of possible concentration distributions, (3) CFD computations on massive parallel supercomputers or (4) tomography. These methods require intensive computations and require ample computer resources. In many cases state-of-the-art research was not directed towards determination of indoor pollutant source locations, but rather to predictions of contaminant dispersion from known locations, which is an inverse problem.

Following such approaches, neural networks (NN) emerged as a tool for rapid concentration forecasting of outdoor environmental contaminants such as, nitrogen oxides (NOx) or sulfur dioxide (SO2). The intention of our present research was to extend applications of NNs to determination of indoor pollutant source locations, making it less computationally intensive. The main research goal was to develop a NN based locator of contaminant sources (LOCS) for fast and reliable determination of pollutant source locations inside a building. Such determination should happen within seconds after receiving real-time contaminant concentration data from building room sensors. For the purpose of NN training, a multi-zone program provided distributions of contaminant concentrations for known pollutant source locations throughout the building. Trained networks had an output indicating pollutant source location based on measured concentrations in different zones of the building.

Three cases based on real building properties demonstrate the applicability of this method to identify contaminant source locations and optimize number and allocation of the required sensors. One of the cases includes experimental validation based upon existing experimental data. Influence of varying outdoor weather parameters was considered to prove applicability of the developed method in real building environments.

Prospective applications of the developed method include: (1) optimizing the contaminant sensor allocation within a building, (2) real-time emergency determination of a pollutant source location during bio-chemical incidents, (3) smoke/carbon-monoxide (CO) source determination in fire protection efforts, (4) determination of a radon source location within a building. Future research intentions are focused on complete automation of the developed method into a one-click user-friendly tool, as well as on integration with the existing sensor networks. Additionally, programs other than multi-zone can supply the training data and make LOCS applicable in variety of other cases, such as outdoor pollution incidents and CBR attacks.

Improvement and Implementation of Simplified Turbulence Model for Airflow Around Multiple Buildings

Airflow around buildings in urban areas including the prediction of airborne contaminants and hazardous materials dispersion around buildings airflow can be successfully resolved using computational fluid dynamics (CFD). However, great variety of flow characteristics and involvement of whole spectrum of turbulent length scale requires the use of carefully derived turbulence models and prescription of boundary conditions.

Current research project comprises both experimental investigation and numerical simulations. The experimental setup consists of group of four buildings at PSU campus. The buildings represent a student dorm complex and are reduced to 1:250 scale. Incoming flow is controlled with an appropriate layout of the spires, trips, and roughness elements at the inlet of the test section of the wind tunnel (left). The building model platform allows the rotation in clockwise and counterclockwise direction to provide different angle of approaching wind (below right). The instantaneous velocity field around model buildings is measured with three dimensional hot wire sensor model 1299A manufactured by TSI. A spatial resolution for collection of velocity time series around a buildings model is presented in the bottom figure. The velocity at the inlet of the buildings is measured at single point four inches upstream at the height of ten inches from the ground by using cross hot wire sensor model X-1243 by same manufacturer. These conditions at single point in the approaching flow are measured simultaneously with velocity measurements around the buildings to provide better control of inlet boundary conditions during the measurements. The signals from hot wire channels are processed and collected on the desktop PC computer by using DISA 55 M01 signal amplifier and conditioner units. The hot wire sensors are previously calibrated against the Pitot static probe in the jet core of small size jet facility with variable speed drive (VSD) fan. The turbulent characteristics of the flow in all three directions are obtained by using look-up table method and procedures developed in MATLAB implementing Delaunay triangulation technique. Experimental results will be used to evaluate the turbulent properties of the flow around the modeled buildings. Thus, a distribution of turbulent length scales around the buildings is crucial for development of new simplified turbulence models.

The second set of activities involves numerical simulations in PHOENICS commercial CFD software, and validation of the developed turbulence model using unsteady Reynolds averaged Navier-Stokes equations (RANS) based on the data extracted from experiments. As a first approximation in the development of simplified turbulence models, the distance to the nearest solid surface in the flow field will be used as the characteristic length scale. The validation of the developed model will be performed against other experimental and field studies available in the literature. The developed turbulence should be easily incorporated in commercially available CFD software such as FLUENT and PHOENICS.

Validation of CFD Calculated Pressure Loads

Computation fluid dynamic programs can quickly simulate fluid-flow properties for a given building structure, but how do the results from CFD compare to other sources of pressure load data? If CFD programs are proven to correspond to past data, it is possible to use CFD in ways never before considered.

During the summer of 2006, results from the CFD program Phoenics 3.6.1 were compared to ASCE 7-02 section 6.0 and measured data from a research project in Canada. In the first comparison, a model of the ASCE 7-02 wind model building structure was built using Phoenics. The image to the right shows the ASCE 7-02 wind load model at the top and the Phoenics model showing the pressure of each surface at the bottom. With the CFD model exactly mimicking the ASCE wind load model, the pressure loads could be compared easily using the charts for the given wind speed and roof pitch. Both the transverse and longitudinal cases of the ASCE 7-02 were examined along with the end zones and the center zone. From the pressure load data and the area of the wall, forces could be calculated for the structure. In this way we could accurately calculate the forces that wind will produce on all surfaces of a building structure. The CFD results will be much more accurate than the ASCE 7-02 results due to assumptions or factors of safety.

In the second comparison, pressure data was collected on a full scale building at Concordia University in Canada. A model was made in Phoenics to resemble the test house. The geometry of the house, the pressure tap locations, and the wind velocity profile matched the onsite conditions as closely as possible. With the model made, the pressure tap readings can be compared to the results from the CFD model by reading the pressure data at each location of the pressure taps. The image on the left shows at the top the house model and the resulting pressure that each surface of the house is experiencing at that particular instant. At the bottom, the picture shows a plan of the model and the fluid-flow around the building. As with the first case, the data collected can be used to determine the forces each wall is experiencing during a wind with that certain velocity and angle.

Computational fluid dynamics programs are very powerful tools and may be able to be used for reasons initially not conceived. Using CFD to find resulting forces for structural reasons is just one way of fully using these programs to their full potential and ultimately helping us better our ways of building structures.

Green Roof R-Values

The thermal performance of green roofs has been studied world wide using three different approaches: field experimentation, numerical studies, and a combination of laboratory or field experiments with numerical models. All the reviewed studies concluded that green roofs can substantially reduce the heat flux from a building roof. However, design engineers do not have a standard calculation procedure or a tool to calculate energy savings from green roofs. Therefore, the goal of this research project is develop an accurate green roof model able to predict the thermal resistance “R” value of a green roof. This “R” values could be implemented in building energy simulation programs commonly used in design practice.

In the summer of 2003 researchers began construction of sample green roofs for the purpose of laboratory testing of their thermal properties. The samples, pictured to the right, are constructed to simulate a green roof on a typical stick built structure. From bottom to top this construction includes plywood sheathing, a vapor barrier, a drainage mat, 4 inches of growing medium, and plants. Two plants, sedum spurium and sedum sarmentosum, were chosen to examine how the choice of flora effects the thermal performance of green roofs. Analysis of previous research and theoretical formulations led researchers to conclude that the R-value will likely be a function of the current weather conditions, the type of plant, and the type of construction. The same analysis showed that the best way to control the wide set of parameters in weather is to establish them in an environmental chamber, thus eliminating the non-steady state condition present in field experimentation.

A new test apparatus was required to utilize the environmental chamber for thermal property tests. Inspired by ASTM standards C177 and C1363, which respectively govern hot plate and hot box tests for thermal properties, researchers developed a cold plate box which would rely on the chamber to serve as a heat source. The apparatus is shown to the right along with two samples of an alternative plastic roof construct currently being tested. The picture to the lower left shows a bank of growing lamps above, which can be suspended at various heights above the green roof samples to simulate varying degrees of solar radiation. The remaining meteorological properties (air temperature, wind velocity, and relative humidity) can be set using the control systems of the chamber in combination with appropriate diffusers. Tests are currently being performed (lower right) and results are expected by 2008.

Strawbale Wall Conductivity

Studies conducted during the spring of 2004 sought to confirm field data gathered on the conductivity of strawbale construction. Strawbale construction is a sustainable building technology which uses straw bales coated in plaster. This technique is inexpensive and easy to construct, often volunteers comprise more than half of the construction teams on these projects. In addition to these qualities, other features have been observed, including good structural strength and large thermal resistances. A full account of the research being done at Penn State on Strawbale can be found at the The American Indian Housing Initiative webpage.

Researchers conducted experiments on two wall samples, roughly 3'x3' (three feet by three feet) and 1' (one foot) thick. Each sample was fitted with thermistors and flux meters to monitor both the temperature and heat flux at various points on and within the samples. After preparing the samples researchers had to prepare the chamber.

The common method for testing heat flux through a material is to generate a temperature difference across the material. One way of accomplishing this is by generating a heat source on one side of the material and a heat sink on the other. The environmental chambers two sections allowed researchers to create such an effect by replacing the windows between the environmental section and climate section with the strawbale walls.

The image to the right shows the final placement of the strawbale walls in the window frame. The blue board around the strawbale walls is rigid insulation, which helps to keep the majority of the heat flowing through the walls. With the walls in place researchers raised the temperature of the climate chamber while lowering the temperature of the environmental chamber. Once the temperature stabilized the researchers gathered data on the heat flow through the walls. Three days after the experiments were complete, the walls were torn down, the windows were replaced, and the next group of researchers began work.

Convection Correlation Development

Building airflow, thermal, and contaminant simulation programs need accurate models for the surface convective boundary conditions. This is especially the case for rooms with:

1. Displacement ventilation (DV) systems
2. Cooled ceiling (CC) systems

With these systems the buoyancy forces at room surfaces significantly affect the airflow pattern and temperature and contaminant distributions. Both systems have recently attracted significant attention from the design community due to their good performance related to air quality and thermal comfort. Therefore, thermal measurements were conducted in the environmental and climate chambers of the BEST facility, in order to developed new surface convection correlations for rooms with these systems.

Convection correlations were developed for steady state conditions, based on measured specific surface heat fluxes, surface temperatures, and air temperature. Using a basic equation for convective heat transfer, the convection coefficients were calculated. Convective heat transfer at internal surfaces was not measured directly, but rather calculated from the temperature measurements. This was accomplished using a heat balance at the internal surfaces as presented in the figure to the left. A system of heat flux meters measured conductive heat flux while the surface temperatures and view factors provided data for calculation of radiative heat flux. Measurements of the temperature difference and corresponding convection coefficient for each surface in each experiment provided a database for the correlation development.

The figure to the right shows the cooled ceiling (CC) panels in the environmental chamber. These hydronic cooling panels removed large sensible cooling loads and provided good thermal comfort in the chamber. For ventilation purposes this system was combined with ventilation systems that provide only fresh air such as dedicated outdoor air systems (DOAS). The figure also shows the installation of a high aspiration diffuser in-between the cooling panels. This diffuser provided a relatively uniform temperature distribution with a small amount of supply air. The measurement results showed that buoyant (natural) convection at the cooled ceiling is the dominant convective mechanism. However, forced convection created by the high aspiration diffuser, depending on the flow rate, increases the convective heat transfer coefficient between 4 and 17%.

In the room with a DV diffuser the major heat transfer from room surfaces to the air appears at the floor surface. Therefore measurement of floor convective heat flux were crucial for the experiments with a DV diffuser. The left side figure shows the DV diffuser and floor of the testing chamber covered with heat flux panels and thermistor sensors. The major parameters that affected the heat flux at the floor surface were: supply air temperature, volume flow rate, and local air temperature. Results show that the correlation based on normalized volume flow rate (ACH) that uses supply temperature as a reference temperature is suitable for heat flux calculation at floor surfaces in rooms with DV.