500 L.D. Zhu et al. the south side in Figs. 4 and 5. Each analogue mechanical anemometer consists of a horizontal component, called RM YOUNG 05106 Horizontal Anemometer, and a vertical component, named as RM YOUNG 27106 Vertical Anemometer. Another two analogue mechanical anemometers (AneM) of horizontal component only were arranged at the level of 11 m above the top of each bridge tower. They are specified as WITPT01 for the Tsing Yi tower and WITET01 for the Ma Wan tower in Fig. 4. The servo accelerometers are of the brand Allied Signal Aerospace Q-Flex QA700. Three different types of arrangement for acceleration measurement were used in the system, namely AccT, AccB and AccU as indicated in Fig. 4, representatively representing Tri-axial measurement (three uni-axial accelerometers assembled orthogonally to each other), Bi-axial measurement (two uni-axial accelerometers assembled perpendicularly to each other) and Uni-axial measurement (by using only one accelerometer to give signal in one prescribed direction). A total of 12 uni-axial accelerometers were located at the four sections of the bridge deck. At each section, there are two accelerometers measuring acceleration in the vertical direction and one accelerometer measuring acceleration in the lateral direction, as shown in Fig.5. A set of three uni-axial accelerometers is located at Ma Wan Anchorage for seismic load measurement. The measurements of wind speed and bridge acceleration response were carried out during the passage of Typhoon Victor over Hong Kong. The sample frequencies were set as 2.56Hz for recording wind speed and 25.6Hz for recording acceleration response, respectively. The recording duration was 7 hours for each channel, from 17:00 to 24:00 on 2 Aug. 1997. Thus, the data number of each time history is 64512 for wind speed and 645120 for bridge acceleration response. By using MATLAB as a platform, some programs were developed to analyse the measured data to obtain mean wind speed, turbulent intensity, integral scale of turbulence, friction velocity, gust factor, wind spectrum, acceleration standard deviation response, and response spectrum of acceleration. The variations of these parameters during the passage of Typhoon Victor were also studied. Each sample (time history) of seven hours duration was evenly divided into seven segments. The segment smooth method and the hamming window were applied in the spectral analysis. For wind spectral analysis, one- hour segment was further divided into 11 sub-segments of 10-minute duration, with an overlapped length of 5 minutes between two neighbouring sub-segments. The 1536 data points in the 10-minute sub-segment were zero-padded to 2048 points to meet the requirement of Fast Fourier Transformation (FFT). The frequency resolution in the wind spectral analysis was thus 0.00175Hz. The spectral analysis of acceleration response was based on one-hour segment, the data point number for FFT and the overlapped length were selected as 8192 and 2.6 minutes, respectively. As a result, one-hour segment of acceleration response could be divided into 21 sub-segments which are averaged and smoothed in the spectral analysis, and the frequency resolution was 0.004375 Hz. MAIN RESULTS: WIND CHARACTERISTICS The wind characteristics around the Tsing Ma Bridge during the passage of Typhoon Victor varied with time due to the change of wind direction and upwind terrain to the Bridge. The wind characteristics also varied with position due to the size of the Bridge and the nature of a typhoon called the heterogeneity of typhoon planetary boundary layer (TPBL). Mean Wind Figures 6 and 7 show variations of 10-minute-averaged mean wind direction and mean wind speed in the horizontal plane with time, respectively. The mean wind was found to be nearly horizontal during the passage of Typhoon Victor at the site of the Tsing Ma Bridge. Figure 6 shows that there is a sudden change of wind direction from north-east to south-west within 20 minutes from 19:50 to 20:10. Correspondingly, the mean wind speed is very small during this period. This is because during this period, Typhoon Victor's eye just crossed over the Bridge. It is also seen from Fig. 6 that the mean wind Wind Response to Tsing Ma Bridge During Typhoon Victor 501 to the Bridge blew from north-east within Region I from 17:00 to 19:50, and from south-west within Region V from 21:00 to 22:00, and from south-west within Region VI from 22:00 to 24:00, respectively. The angles between the mean wind measured at the top of the tower and the longitudinal axis of the bridge were about 20 ~ , 34 ~ and 52 ~ , correspondingly. During the period of 20:10 to 21:00, the mean wind direction of Typhoon Victor at the site of Tsing Ma Bridge was unstable and varied from Region II to Region IV due to the landfall on the mountain areas. The maximum 10-minute mean wind speeds were measured as 12.9m/s at the deck level and 16rn/s at the tower-top level before the Typhoon Victor crossed the Bridge. After the crossing, they became, respectively, ll.7m/s and 14.2m/s during 20:10 to 21:00, 18.5m/s and 21.1m/s during 21:00 to 22:00, and 17.4rn/s and 23.3rn/s during 22:00 to 24:00. The maximum hourly mean wind speeds before the crossing were 10.1m/s at the deck level and 13.9m/s at the tower-top level. After the crossing, they became, respectively, 7.9m/s and 9.3m/s in the duration of 20:00 to 21:00, 14.9m/s and 17.8m/s in the duration of 21:00 to 22:00, and 15.7m/s and 21.2m/s in the duration of 22:00 to 24:00. The highest 10- minute and one-hour mean wind speeds at the tower-top level occurred between 22:00 and 23:00. Figure 6: Variation of 10min mean wind direction at WITPT01 Figure 7: Variation of 10min mean wind speed at WITPT01 Figure 8: Variation of gust factor with gust duration at WlTJS01 The mean wind speed profile during a typhoon is not well known yet. It also cannot be exactly explored this time, for only two level wind speeds were available. However, by fitting two level mean wind speeds to the power law mean wind profile, it was found that the mean value of exponent for the power law was 0.324 when wind blew from north-east within Region I and 0.199 when wind blew from south- west within Region V. These values are very close to those specified in the Hong Kong Wind Code (1983) for the build up terrain and general terrain respectively. Turbulence Intensity and Gust Factor Compared with seasonal trade winds, the turbulence intensity of wind due to a typhoon is relatively higher. The measured largest hourly mean longitudinal turbulence intensity was about 33% at the deck level and 27% at the top level. The measured largest hourly mean lateral turbulence intensity was about 27% at the deck level and 23% at the top level. The corresponding value for vertical turbulence intensity was about 25% at the deck level. With respect to the gust factor based on the hourly mean wind speed, it is found that for a given longitudinal turbulence intensity, the factor is approximately proportional to the logarithm of gust duration (see Fig. 8). The factor is also almost proportional to longitudinal turbulence intensity for a given gust duration. Thus, by best fitting the measured data, the following empirical formula is obtained for the estimation of the gust factor during Typhoon Victor. G(T, Iu) : 1- 0.5377(Iu) 1~ In(T/3600) (1) where, T is the gust duration in second; I u is the longitudinal turbulence intensity. 502 Wind Auto Spectra L.D. Zhu e t al. Friction velocity u, was estimated through the horizontal shear stress (Tieleman and Mullins 1980) at the deck level. The friction velocities were found to be 1.23m/s, 1.09m/s and 0.86m/s for Regions I, V and VI, respectively. The reduced auto spectra of three components of fluctuating wind (nSu/U, 2 , nS v/u 2 , nS w/u, 2 ) varied strongly with the time due to the change of wind direction and upwind terrain and also with the height of the anemometer position. However, the slopes of all auto spectra were approximately equal to -5/3 in the reduced frequency range f > 0.3 for nSu/u 2 and f > 0.6 - 0.9 for nSv/u, 2 and nS w/u, 2 . The reduced frequency f is equal to nz/U(z), where U(z) is the mean speed at height z and n is the frequency in Hertz. The auto spectra can be fitted using non-linear least square method with the following objective function. nSa/u. 2 = a f/0 +bfl/m) 5m/3 (2) where the subscript a of S can be u,v, or w; and a, b, m are the parameters to be fitted. The results showed that these parameters scatter in a wide range, depending on the upwind terrain and the position of anemometers. Figures 9 to 11 display the longitudinal, lateral, and vertical wind spectra measured from WITJS01 during the period of 22:00 to 23:00 on 2 August 1997. The corresponding fitted curves together with the von Karman spectra, Kaimal spectra and Simiu spectra are also plotted in these figures (Morfiadakis et al 1995, Kaimal et al. 1972, Simiu & Scanlan 1996). It is seen that the spectra using Eq. 2 can fit the measured spectral data well in both low and high frequency regions. The von Karman spectra, using measured integral scales, fit the measured spectral data better than Kaimal and Simiu spectra, especially in the low frequency region. Figure 9" nS u/u 2 at WITJS01 Figure 10: nS v/u2. at WITJS01 Figure 11" nS w/u 2 at WITJS01 MAIN RESULTS: ACCELERATION RESPONSE Standard Deviation and Peak Factor Due to the limitation of space, only the acceleration responses of the Bridge at the mid-span after the crossing of Typhoon Victor are presented. Figure 12 illustrates the time histories of lateral, vertical and torsional acceleration responses during the period of 22:00 to 23:00. Figure 13 shows variations of the aL aV aT standard deviations (0-60,0-60,0-60) of lateral, vertical and torsional accelerations with time in 10-minute aL aV and aT interval. The maximum values of 0-60,0-60 0-60 were found to be 0.588cm/s 2 , 3.082 cm/s 2 and 0.0010 rad/s 2 , occurring at about 21:35, 23:35 and 22:45, respectively. Figure 14 shows the relationship of O"60aL, (5.60aV , 0-60aT with the 10 minutes mean wind speed (U6o). It can be seen that the vertical and aV and aT increase almost proportionally to the cube of the mean wind torsional acceleration responses c60 0-60 Wind Response to Tsing Ma Bridge During Typhoon Victor 503 aL increases almost proportionally to the square of the speed U60 while the lateral acceleration response ~60 mean wind speed U60. The maximum values of the one-hour standard deviation acceleration responses aL aV and aT during 22:00"~23:00 were 0.468 cm/s 2 2.050cm/s 2 and 0.0007 rad/s 2 respectively. CY360 , l~Y360 I~Y360 , , The maximum peak accelerations during the period of 22:00 to 23:00 were 1.95 cm/s 2 , 9.78 cm/s 2 and 0.0034 rad/s 2 . In terms of one-hour duration, the averages of the measured peak factors for seven hour duration were 4.86, 4.29 and 5.73, respectively, for lateral, vertical and torsional acceleration responses, but in terms of 10-minute duration they became 3.71, 3.04 and 3.51, respectively. Probability analysis of acceleration peak factor was also performed. It was found that the peak factor distributions comply with the Gaussian distribution approximately. Figure 13" Variation of standard deviation of acceleration Figure 12: Responses of lateral, vertical and torsional accelerations at the mid span (22:00 to 23:00) Figure 14: Acceleration response via mean wind speed Figure 15" Acceleration spectra of the Bridge at mid-span (22:00 to 23:00) Acceleration Response Spectra Figure 15 illustrates the auto spectra of the lateral, vertical and torsional acceleration responses at the mid-span of the Bridge for the period of 22:00 to 23:00. The first two peak frequencies identified from these spectra were 0.0688Hz (0.068Hz)[0.069Hz] and 0.2656Hz (0.285Hz)[0.297Hz] for lateral 504 L.D. Zhu et al. acceleration, and 0.2656Hz (0.271Hz)[0.267Hz] and 0.4844Hz (0.475Hz) for torsional acceleration. The first three peak frequencies identified from the vertical spectrum were 0.1375Hz (0.137Hz) [0.139Hz], 0.1813Hz (0.189Hz)[0.184Hz] and 0.325 Hz (0.325Hz)[0.327Hz]. Compared with the numbers in the above parenthesises and square brackets that were obtained by Xu et al (1997) from the eigenvalue analysis and the ambient vibration measurement respectively, one can see that three sets of the natural frequency results are very close. The relative difference of the lateral frequencies is less than 11% and that of the torsional frequencies is less than 2% whilst that of the vertical frequencies is less than 4%. Furthermore, these frequencies were found to remain almost constant during the passage of Typhoon Victor in spite of the variation of the wind speed, wind direction, and upwind terrain. CONCLUSIONS The recorded wind and structural response data were analysed in this paper for evaluating wind characteristics and acceleration response of the Bridge. The results show that during Typhoon Victor, both mean and turbulent wind characteristics varied considerably due to the change of wind direction and the upwind terrain. Turbulence intensities and gust factors measured during Typhoon Victor were higher than those due to seasonal trade winds. An empirical formula for gust factor as a function of turbulence intensity and time duration was also provided based on the measured results. It was confirmed that the wind excitation mechanism of the Bridge in the lateral direction was different from that in the vertical direction or the rotation. The alongwind acceleration response of the Bridge was approximately proportional to mean wind speed square while the vertical acceleration and torsional angular acceleration were almost proportional to mean wind speed cubic. Furthermore, the natural frequencies identified from the acceleration response spectra were consistent with those obtained from the ambient vibration measurement or the numerical analysis carried out before. ACKNOWLEDGMENTS The work described in this paper was supported by both the Hong Kong Polytechnic University through a studentship to the first writer and the Research Grants Council of Hong Kong through a grant to the second writer (Project No. PolyU 5027/98E). Any opinions and conclusions presented in this paper are entirely those of the writers. REFERENCES Code of Practice on Wind Effects: Hong Kong-1983, Building Department, Hong Kong Kaimal J.C., Wyngaard J.C., Izumi Y. and Cote R. (1972), Spectral characteristics of surface-layer turbulence, J. Royal Meteorol. Soc., Vol. 98:132-148. Lau, C.K., Wong K.Y. and Chan K.W.Y.(1998), Preliminary mornitoring results of Tsing Ma Bridge, The 14 th National Conference on Bridge Engineering, Shanghai, Vol.2:730-740 Li P.W., Poon H.T. and Lai S.T. (1998), Observational study of Typhoon Victor (9712) during its passage over Hong Kong, 12 th Guangdong-HongKong-Macau Seminar on Hazardous Weather, Hong Kong (in Chinese). Morfiadakis E.E., Glinou G.L. and Koulouvari M.J. (1995), The suitability of the von Karman spectrum for the structure of turbulence in a complex terrain wind farm, J. Wind Eng. Ind. Aerodyn. Vol 62:237- 257 Simiu E.& Scanlan R.H. (1996), Wind effects on structures, John Wiley & Sons, INC, New York. Tieleman H.W., Mullins S.E. (1980), The structure of moderately strong winds at a Mid-Atlantic coastal site (below 75m), Proceedings of Fifth International Conference on Wind Engineering, Pergamon Press, Oxford, Vol 1:145-159 Xu Y.L., Ko J.M. and Zhang W.S. (1997), Vibration studies of Tsing Ma Suspension Bridge, J. Bridge Eng. ASCE, Vol 2:149-156. STRUCTURAL PERFORMANCE MEASUREMENT AND DESIGN PARAMETER VALIDATION FOR KAP SHUI MUN BRIDGE C.K. Lau ~ W.P. Mak ~ K.Y. Wong ~ Deputy Director Chief Engineer Senior Engineer K.L. Man I W.Y. Chan ~ K.F. Wong z Engineer Engineer System Analyst tHighways Department, The Government of the Hong Kong Special Administrative Region /E&M Section, Tsing Ma Management Limited (The Operator of Tsing Ma Control Area) ABSTRACT A structural health monitoring system has been operating on the Kap Shui Mun (cable-stayed) Bridge, as part of a strategic road and rail Link in Hong Kong, in order to monitoring and evaluating structural performance of the bridge structures. Valuable field measurement data of the bridge responses under various environmental and traffic conditions are being obtained to establish a baseline reference. This paper briefly describes the system of data acquisition and presents some findings in the evaluation of the structural health conditions of the bridge with reference to the design values. KEYWORDS cable-stayed bridge, structural health monitoring, wind, temperature, traffic load, bridge response INTRODUCTION Kap Shui Mun Bridge (KSMB), forming part of the Lantau Link in Hong Kong, is the first road and rail cable-stayed bridge ever built in Hong Kong. Since its opening in May 1997, this strategic crossing has been serving as the only road and rail link to Lantau Island and the Hong Kong International Airport (Figure 1). The 430m main span of the double-deck cable-stayed bridge has adopted an innovative steel-concrete composite design in its main span and coupled with concrete side spans at the two ends (Figure 2). As a pioneer in Hong Kong to monitor structural performance of bridges by field measurement, Highways Department (HyD) has installed a permanent On-structure Instrumentation System (OSIS) on the bridge. This OSIS is used to collected the measured information for structural health monitoring and design data validation works. 505 506 C.K. Lau et al. OVERVIEW OF THE ON-STRUCTURE INSTRUMENTATION SYSTEM The OSIS for KSMB is comprised of a total of 270 sensors of different types permanently installed on the bridge (Figure 2), such as anemometers, accelerometers, level sensing system, strain gauges, temperature sensors, displacement transducers and weigh-in-motion sensors. Streams of data signal. are continuously transmitted from these sensory systems to the controlling computer system (CLFC) located in the Bridge Monitoring Room at 1 ~t Floor of the Tsing Yi Administration Building. In the transmission network, there are acquisition computers which are housed inside air-conditioned cabinets of the two Outstation Units located inside the lower deck of the bridge. Analogue and digital signals are collected into these acquisition computers via 313 numbers of signal channels and undergo signal conversion with appropriate signal conditioning and amplification devices before transmitting via a token-ring network over fiber optic cables to CLFC in Bridge Monitoring Room. The measured data are then processed by tailor-made programmes, operating in Unix platform, for real-time display of strategic data as well as subsequent data analysis and archiving. WIND LOAD MONITORING The cable-stayed bridge was designed to withstand an hourly mean wind speed of 50 rn/s and a maximum 3-second gust wind speed of 80 m/s. The design wind speeds under different live load conditions on deck structure are given in Table 1 below. TABLE 1 DESIGN WIND SPEEDS FOR BRIDGE STRUCTURES Live Load Conditions on Deck Structure Without Highway and Railway Live Loads With Combined Highway & Railway Live Loads With Railway Live Load Mean Hourly Wind Speed at Deck Level 50 m/s 25 m/s 28 m/s Max. Wind Gust Speed (rru's) Horizontal Wind loaded length <_20m 100m 600m 1000m 80 72 65 63 44 38 34 33 50 43 39 38 Min. Gust Wind Speed 50m/s 25 m/s 28 m/s The above wind design parameters were derived from available statistical wind records obtained at Waglan Island which is situated about 30 kilometers from the bridge site. Different topography would have influence on the wind parameters. A database established from the wind measurement data is used to verify the design wind parameters. Two anemometers of ultrasonic type are installed at deck level of around 59mPD and positioned at mid-span on outrigger trusses on both sides of the deck. Wind rose diagram is derived from the wind measurement, showing wind speeds, wind directions and frequencies of occurrence, as shown in Figure 3 for a two-year period since the opening of the bridge. The measurement record illustrates a relatively mild condition of wind speed in the past two years. TEMPERATURE MONITORING The design temperature data specified for KSMB was generally adopted from Structural Design Manual [3]. These design values were derived by theoretical approach based on experimental data previously used in deriving the temperature data for design of concrete bridges under Hong Kong climatic conditions. With temperature measurement in place, the range of design temperature parameters specific for this steel composite bridge can be verified. Temperature sensors are installed in various locations of the bridge. Temperatures are recorded in structural steel beams at top and Structural Performance Measurement for Kap Shui Mun Bridge 507 bottom deck levels, steel cladding on fascia structures, concrete tower legs, road asphalt on upper carriageways. Air temperatures above, inside and below the deck are also recorded. Differential temperature within the bridge as induced by solar radiation and re-radiation can then be identified. The theoretical effective temperature, with appropriate weighting of cross-sectional areas, can also be verified by calculation. The design range of effective bridge temperature for different structural components are tabulated in Table 2 below for information. TABLE 2 DESIGN EFFECTIVE BRIDGE TEMPERATURE Structural Components Design Effective Bridge Teml Steel Composite Main Span * Concrete Side Span * 36 ~ Stay Cables 50 ~ Concrete Towers 36 ~ Note : * Thickness of deck surfacing is 100mm. perature for 120-year return period Maximum Minimum 40 ~ 0 ~ 0~ -2 ~ -1 ~ Figure 4 illustrates the variation of temperature recorded in bridge-deck over a 12-month period. Both shade air temperature and calculated effective bridge temperatures exhibit a typical seasonal variation. During the period, the maximum and minimum effective bridge temperatures of 36 ~ and 8 ~ were recorded respectively. It demonstrates that temperature variation in bridge-deck has remained within the design temperature range of 40 ~ and 0 ~ (for a 120-year return period). Since the alignment of the bridge orientates in a North-East to South-West direction at 55 ~ from the North, it is anticipated that the solar radiation along the locus path from sunrise to sunset will induce a thermal gradient across the hollow section of the twin tower legs. Platinum resistance thermometers installed at 48m above deck level are used to monitor the temperature of concrete tower legs. Typical variations of temperature gradient across the hollow section during a hot summer period of 72 hours are shown in Figures 5 and 6. The temperatures on east face of tower leg build up quicker in the morning while west face is still under shade. The temperatures on west face begin catching up after mid-day when the sun is high up. There is also a gradual reversal of temperature gradient in late afternoon between outside (sensor H) and inside face (sensor A) of concrete tower wall as the shade temperature inside tower legs remain reasonably constant throughout the day. The difference in temperatures between the two faces of wall has reached a maximum of 5 ~ which is still within the permissible range. The corresponding ambient temperature is also plotted for comparison. Incidentally, there was a shower in the morning on the first day which caused the temperature drop. VEHICULAR TRAFFIC LOAD MONITORING In the design of highway live loads on bridges, various combinations of vehicular loading patterns are explored to encompass the most adverse configuration of vehicular traffic loads. The vehicular traffic loads and associated load factors are derived by statistical simulation on the basis of projected future growth of traffic volume and the proportion of heavy goods vehicle population amongst other classes of vehicles. The distribution of different classes of vehicles in traffic jams is in fact the governing factor in deriving the vehicular traffic loads. The loaded length used in design also depends on frequency of daily traffic jams, locations of traffic jams, duration and distribution of vehicular types as well as traffic flow during traffic jams. For Lantau Link, a weigh-in-motion (WIM) sensory system is installed on both bounds of the carriageways near the Lantau Toll Plaza. The WIM system (bending 508 C.K. Lau et al. plate type) is used to record vehicle count, to measure axle weight and travelling speed and to identify type of vehicles crossing the bridges. A database is being established from these statistical records to verify the population of heavy goods vehicles. Correlation can then be made with the design HA lane factor of 3.6 adopted in highway live load design. Figure 7 is a monthly average statistics of daily vehicle count and proportion of different types of goods vehicles in traffic population. Population of goods vehicle is demonstrated to be around 34%, which is well below the design value of 60%. DYNAMIC RESPONSE MONITORING As a means to monitor the global dynamic characteristics of the bridge, accelerometers are installed on the deck to record its ambient vibration status. From accelerometer data, natural frequencies are derived. Mode shapes and modal damping values of the deck are subsequently determined. Since there are only three uni-axial accelerometers permanently installed at mid-span section of the deck, additional measurement points were established on the deck by setting up portable measuring equipment at four sections of the main span to obtain additional deck vibration data over short periods of time. The measured and computed frequencies of the deck are in Table 3 below. TABLE 3 COMPARISON OF COMPUTED AND MEASURED NATURAL FREQUENCIES Mode No. Computed Frequencies (Hz) Consultant[ HyD 1 0.378 0.41 2 0.509 0.58 3 0.645 0.93 4 0.955 1.51 5 1.199 1.74 6 1.296 2.81 7 1.483 2.88 1 0.383 0.49 2 0.876 1.15 3 1.035 2.45 4 1.298 3.06 5 N/A 3.39 1 0.676 0.77 2 1.137 1.62 3 1.442 2.18 4 N/A 2.69 5 N/A 3.35 Measured Type of Mode Shape Frequencies (Hz) University HyD ! (A) For Deck Vertical-dominant Modes 0.39 0.39 Vertical 1 i i 0.66 N/A Towers i i 1.07 0.68 Vertical 2 i i 1.54 1.05 Vertical 3 i | 1.81 N/A Towers 2.71 ' 1.35 ' Vertical 4 (Towers in phase) 3.08 1.53 . Vertical 5 (Towers out of phase) (B) For I)eck Lateral-dominant Modes 0.49 0.41 Lateral 1 i i , 1.25 0.90 Lateral 2, Towers, Torsion 2 | | 2.12 N/A Torsion 2, Lateral 2, Towers i | 2.93 1.31 Ma Wan Side-span Lateral, Tower ! i 3.20 N/A (C) For l)eck Torsion-dominant Modes 0.83 0.81 Torsion 1 I I 1.39 1.18 Torsion 2 i ! 1.90 1.77 Torsion 3 I I 2.56 N/A Torsion 4 m j 3.39 N/A : Torsion 5 In above table, the natural frequencies previously computed by Consultant by means of computer model simulation are presented for comparison. With the limited number of measurement points on the deck, the mode shapes of towers and side spans, coupling modes of deck, towers and stay cables cannot be identified. A university was once commissioned to conduct a field measurement of ambient vibration of the cable-stayed bridge with a total of 34 measurement points (18 nos. along the deck, 4 Structural Performance Measurement for Kap Shui Mun Bridge 509 nos. at each tower and 8 nos. in stay cables). Their corresponding measured values are also tabulated for comparison. However, the two measurement results do not provide sufficient agreement to ascertain the correctness of the measured natural frequencies. Further measurement would be required to verify the values. In structural design of this road and rail bridge, there are certain criteria stipulated by railway operation to ensure acceptable levels of comfort for users of the railway. Some of these are more stringent than those required for highway traffic. One of them is the limiting of vertical acceleration of the deck. The measured vertical acceleration of the deck has demonstrated that the deck dynamic response is far below the design normal maximum value of 0.05g and only occasionally approaches the threshold of desirable upper limit of 0.03g as shown in Figure 8. STRAIN AND STRESS MONITORING As the bridge is constantly responding to the ever-changing environmental loading and fluctuating live load in addition to static permanent loads, various structural components of the bridge are subject to varying strain and stress conditions. Monitoring of strain in certain critical structural components by strain gauges and level sensing system can provide a health histories of the bridge and establish baseline references for damage detection/assessment. During June 1999, Typhoon Maggie approached Hong Kong and there was a change in direction of her path. Records of structural responses in some of the instrumented structural components are extracted for a 60-minute period as presented in Figure 9 to 11. It is noticed that the change of wind incidences has induced a corresponding changes in bridge responses, causing an observable reduction in both the vertical oscillation of the deck and variation in strain/stress levels recorded in the sensory systems. Nevertheless, the strain/stress levels were well below permissible design values. CONCLUSION The structural health monitoring system enable continuous collection of field measurement data, reflecting different bridge responses under various types of environmental and applied loads. A health history with baseline references can be established over times. Current structural health monitoring work shows that the bridge is in a healthy condition. ACKNOWLEDGEMENT The authors express their thanks to Director of Highways, Mr. K.S. Leung, for permission to publish this paper. Any opinions expressed or conclusions reached in the text are entirely those of the authors. REFERENCES 1. Lau, C.K. and Wong, K.Y., "Design, Construction and Monitoring of the Three Key Cable-supported Bridges in Hong Kong", Proceedings of" the Fourth International Conference on Structures in the New Millennium", 3-5 September 1997 in Hong Kong, A.A. Balkema, Rotterdam, Netherlands. 2. Lantau Fixed Crossing Project Management Office, Highways Department, "Structural Health Monitoring System", Highway Contract No. HY/93/09 - Electrical and Mechanical Services in Lantau Fixed Crossing, The Hong Kong Government, 1993. 3. Highwaysn Department, "Structures Design Manual", The Government of Hong Kong Special Administrative Region, 1997. . sub-segments of 10-minute duration, with an overlapped length of 5 minutes between two neighbouring sub-segments. The 1536 data points in the 10-minute sub-segment were zero-padded to 2048 points to. standard deviations ( 0-6 0, 0-6 0, 0-6 0) of lateral, vertical and torsional accelerations with time in 10-minute aL aV and aT interval. The maximum values of 0-6 0, 0-6 0 0-6 0 were found to be 0.588cm/s. cable-stayed bridge ever built in Hong Kong. Since its opening in May 1997, this strategic crossing has been serving as the only road and rail link to Lantau Island and the Hong Kong International