1、4702 A Preliminary Investigation on the Use of Ultra-Wideband Radar for Moisture Detection in Building Envelopes William M. Healy Member ASHRAE ABSTRACT A preliminary investigation has been carried out to deter- mine the potential for using ultra-wideband (UWB) radar to determine the moisture level
2、within building envelopes. Radio waves are aflected by moisture content because their reflection from the surface of a material depends upon the dielectric constant of that material, aproperty with a strong dependence on the moisture content. UWB radar holds the potential for gaininggreater informat
3、ion from a wall than can be obtained by conventional radar because of the large frequency range covered by emitted signals. Tests on small samples of oriented strand board (OSB), pine, and gypsum board have shown that the energyreflectedfrom thesamples increases with increasing moisture content. Fur
4、ther investigations were carried out on a simulated wall consisting ofpanels of gypsum board, insu- lation, and OSB sheathing that were conditioned to varying moisture contents. Algorithms have been utilized that enable the user to separately identifl the moisture levels of the indi- vidual layers i
5、n the wall. The last part of the study joined the radar unit with mapping software to create three-dimensional images of the moisture condition of a wall. These results may lead to a new technique to nondestructively map the hygro- thermal state of the building envelope. INTRODUCTION Moisture proble
6、ms in buildings have received an increas- ing amount of interest as durability and mold issues grab the publics attention. Recent litigation concerning mold has heightened the need for studies of moisture movement in building envelopes to determine best practices for preventing damage. To this end,
7、various studies have taken place and numerous tools have become available to predict moisture Eric van Doorn damage. Computer models have been developed (Burch et al. 1995; Karagiozis et al. 2001) that simulate the movement of water vapor or liquid water through a building system to predict areas of
8、 unwanted moisture accumulation, and hand- books are available that provide guidelines for the construc- tion industry (Lstiburek and Carmody 1996; Trechsel 1994). Additionally, a wide range of laboratory and field studies has been carried out to determine the hygrothermal behavior of building envel
9、opes. Attention must be given not only to the initial construction but to the continuing operation of the building in order to prevent mold growth or structural decay caused by moisture accumulation. In order to determine the moisture levels within a building effectively, suitable sensors are needed
10、 to measure the mois- ture content of building materials. Such measurements are necessary for validating computer simulations of moisture transport, for laboratory experiments, and for field studies. The standard method of determining the moisture content of materials is through gravimetry, where th
11、e mass of a moist specimen is compared to its dry mass (ASTM 1992). Such a technique, however, is often not practical for in situ measure- ments. Other technology for making in situ measurements is limited and may not meet all current needs. Several reviews have discussed the technologies available
12、for making measurements of the moisture content of materials within a wall or roof (Healy 2003; Derome et al. 2001; Ten Wolde and Courville 1985). Currently, the most popular methods for making in situ measurements ofmoisture content are electrical resistance measurements of the material with pin pr
13、obes and capacitance techniques. The electrical resistance of wood is highly dependent upon its moisture content, though it is also dependent upon the species and the temperature of the wood. William Healy is a mechanical engineer in the Building and Fire Research Laboratory, National Institute of S
14、tandards and Technology, Gaith- ersburg, Md. Eric van Doorn is a senior scientist at Intelligent Automation, Inc., Rockville, Md. 02004 ASHRAE. 95 Electrical resistance measurements provide a fairly accurate measure ofthe moisture content given proper processing algo- rithms, but they are susceptibl
15、e to drift and noise. Further- more, insertion of the pin probes that supply the voltage over which the resistance is measured is a destructive process. Determining the moisture content below the walls surface requires either installation of pins before construction or alter- ation of an existing wa
16、ll, making for a very costly set of measurements. Meters that measure the capacitance of a mate- rial can be held against a wall to make the measurement of the moisture content within the wall. These measurements are also dependent upon the material being measured and the temperature of the material
17、. While the measurement is nonde- structive, its difise nature does not give precise spatial infor- mation concerning the moisture problem. Surface condensation can also seriously impact accuracy of the read- ings. Another technique to determine the moisture content within a building material involv
18、es the use of humidity sensors. With knowledge of the humidity adjacent to a mate- rial, sorption isotherms can be used to estimate the moisture content of that material. Implementation of this technique can be easier than the others mentioned because of the maturity of the humidity sensor market, b
19、ut the use of sorption isotherms in this application is questionable because they represent the equilibrium states. Achieving these states requires significant time to attain in buildings, so fluctuations in the humidity will not correlate immediately with changes in the moisture content. The review
20、s of sensors all concluded that improved techniques for determining the moisture content within a building envelope would aid building practitioners in detect- ing moisture problems and in evaluating different construction techniques. An emerging technology that can be used to determine the moisture
21、 content within a wall is ultra-wideband (WB) radar. This technology has received increasing attention as a technique for wireless communications and object detection. This paper will discuss experiments that were performed to demonstrate the potential for this technology to determine the moisture c
22、ontent of different layers within a wall nondestruc- tively. Initial investigations to prove the concept will be presented, and experiments on a small sample wall section will then be discussed. Finally, the UWB radar has been combined with mapping software to provide three-dimensional images of moi
23、sture accumulation in a wall cavity, OVERVIEW OF ULTRA-WIDEBAND RADAR Traditionally, the use of radio waves for moisture measurements has been confined to soil measurements and, more recently, for determining the moisture level in agricul- tural products. The key reason for the sensitivity of radio
24、waves to moisture lies in the large difference between the dielectric constant of water (E, = 81) and that of dry soil and other porous materials such as wood (E, x 5). The dielectric constant affects the propagation of radio waves in several ways. First, the speed of propagation of electromagnetic
25、waves, c, through a solid is dependent upon the dielectric constant as described by Equation 1. where = the magnetic permeability of free space ,ur = the relative magnetic permeability of the medium to that of free space eo = the permittivity of free space E, = the dielectic constant of the medium T
26、he second way in which moisture affects the propaga- tion of radio waves through a solid is through reflection at the interface between the surface of the solid and the adjacent material. The reflection at the interface between materials 1 and 2 is governed by the reflection coefficient, T,*. where
27、c1 and c2 are the speed of light in materials 1 and 2, respectively, as described by Equation 1. This number provides the ratio of the amplitude of the reflected waves to that of the incoming waves. Since c1 and c2 depend upon the dielectric coefficient, the reflection coefficient also depends upon
28、the dielectric coefficient. The dielectric coefficient affects attenuation of the electromagnetic waves as well, though this trait will not be used in this sensing application. It should also be noted that the dielectric constant is a function of frequency and temperature. Adjustments may be needed
29、to any measurements made at temperatures differing greatly from that at which a material is calibrated. The fact that the dielectric constant varies with frequency is actually one of the advantages envisioned for the use of ultra-wideband radar and will be discussed shortly. Conventional radar is a
30、narrowband device, Le., it uses a single frequency in the radio spectrum. The use of multiple frequencies, however, yields more information, in principle, than a single frequency because the dielectric constant of moisture is a function of frequency. More information can therefore be obtained by emi
31、tting a broadband signal toward a surface and detecting the refiection of various frequencies. Several hundred MHz of bandwidth can be achieved by sweeping frequencies, but a time domain approach or pulsed signal is needed to achieve ultra-wideband (UWB). In fact, the Fourier transformation of an in
32、finitely short pulse in the time domain shows that the pulse corresponds to a signal of infinite bandwidth in the frequency domain. One such implementation of the time domain approach is time domain reflectometry (TDR). TDR has been used to measure soil moisture by inject- ing a pulse having a durat
33、ion of several ns into a wire embed- ded in the soil and analyzing the reflected signal (Topp et al. 1980). The moisture level in the soil affects its dielectric constant, and the propagation of the pulse is delayed because of that change. Such broadband techniques have not gained 96 ASHRAE Transact
34、ions: Research O 0.2 0.4 0.6 0.8 1 Time (ns) I Figure 1 A Gaussian monocycle pulse and its specti-um. -0.6 wide acceptance, though, because of the fast timescales, the need for equipment to generate high bandwidth signals, and the limited number of units. These issues have caused the elec- tronics t
35、o remain complicated and expensive. In the last five years, however, communications companies have invested heavily in developing chip-based electronics for UWB radios. These developments have led to prototype UWB devices that are available for testing. With a recent ruling by the Federal Communicat
36、ions Commission that permits the use of UWB for wireless communications in the United States (FCC 2002), sales volumes are expected to rise strongly over the next two years, and prices are expected to drop to as low as $50 per chip set. This drop in cost opens the door to alternative applications, s
37、uch as moisture sensing in building envelopes. This topic is the subject of this paper. I l l YI I _I_ , I I UWB Basics At the most fundamental level, conventional radar and communication systems modulate a carrier frequency in order to transmit information. In UWB technology, however, another mecha
38、nism is used for transmitting information. There is no carrier frequency in UWB. Instead, very short pulses of energy are emitted. As mentioned previously, a perfect impulse in the time domain has an infinite bandwidth in the frequency domain. Energy is therefore spread over all frequencies. In prac
39、tical implementations, pulses with widths of several nanoseconds to 200 picoseconds are generated with resulting bandwidths of several GHz. The UWB hardware used in this study is a commercially available product that emits ultra-short “Gaussian” monocycles, as displayed in Figure 1, in both the time
40、 domain and the frequency domain. These pulses are emitted in tightly controlled intervals, gener- ating a train of signals. The impulse receiver directly converts the received RF signal into a base-band digital or an analog output signal. An analog cross-correlator coherently converts the electroma
41、gnetic pulse train to a base-band signai in one stage. No intermediate frequency stage is needed, thereby greatly reducing the complexity of the unit. A Pulse Train For radar and communication, impulse systems use long sequences of pulses rather than single pulses. The use of a train of pulses is es
42、sential because each pulse is generated at an Q) 5 O4 5 02 Y $ 3 -02 O 2 d here, only the summation of the square of the voltages was used. The limited results at least show some dependency of the radar return to moisture content variations in building materials. Tests on Larger Panels in a Simulate
43、d Wall The next tests were performed on larger panels of gypsum board and OSB measuring 0.61 m x 0.91 m x 1.27 cm (2 ft x 3 ft x !h in.). The moisture content of the panels was deter- mined gravimetrically. The OSB samples were dried in an oven set at 103C (217F) until the change in mass was less th
44、an O. 1 %. The gypsum board was initially dried in an oven set at 45OC (1 13F) and then placed in a desiccating chamber to complete the drying process, as suggested by ASTM Test Method C472 (ASTM 1999). After obtaining dry masses of the samples, the panels were placed in chambers containing saturate
45、d salt solutions that created relative humidities of 33%, 58%, and 84%. Sensors placed in these chambers indicated that the humidity within the chamber fluctuated by approxi- mately f10% from the desired conditions because of infiltra- ASHRAE Transactions: Research tion of room air into the chambers
46、. The panels were weighed periodically to determine their moisture contents and were used as part of a simulated wall, as will be discussed next. Moisture contents are determined with an uncertainty of +0.5%. The measured moisture contents of the three OSB panels at the time of testing were 5.7%, 8.
47、3%, and 13%, and those of the gypsum board panels were 0.63%, 0.73%, and 1.1%. Simulated Wall A set of tests was performed on a simulated wall section. The purpose of these tests was to investigate the ability of the radar to detect moisture contents ofdifferent layers with a wall. The wall section
48、was composed of 8.9 cm (3.5 in.) ofR-13 fiber- glass insulation sandwiched between a 1.27 cm (0.5 in.) sheet of gypsum board and a 1.27 cm (0.5 in.) layer of oriented strand board. The insulation had kraft paper that faced the gypsum board. Panels of these materials measuring 0.61 m (2 ft) wide by 0
49、.9 1 m (3 ft) tall were placed within a frame constructed of 2-by- 4s of sugar pine. The rig was constructed so that panels of vary- ing moisture content could be interchanged to obtain a wide range of moisture conditions. Moisture content measurements were not made on the insulation because of the low moisture storage capacity of the fiberglass. Their moisture levels will instead be identified by the relative humidity at which they were stored (%TO, 58%, 84%). The OSB and gypsum board panels were those previously discussed that were conditioned in tubs of constant relative humidity. To
