Industrial Sponsorship

Dynamic Performance Investigation of Base Isolated Structures

By Ather K. Sharif

9.1 Introduction

This case study examines the dynamic performance of a base isolated building known as Eland House, which is partly built over the Victoria Line, in London. The performance of this base isolated building is compared with that of a neighbouring un-isolated structure of similar size and proximity to the tunnel. This comparison is used to provide an indication of insertion loss.

9.2 The Site

A key plan in Figure 9.1(a) shows northbound and southbound tunnels of the Victoria Line merge into a single crossover tunnel, immediately beneath the southeast corner of the base isolated Eland House site. The single crossover tunnel is seen to continue under the western edge of an un-isolated building known as Watney House. The site is located close to the station, which explains the presence of the crossover tunnel. The crown of the tunnel is at a depth of 15m from pavement level. The isolated and un-isolated structures are shown in plan in Figure 9.1(b) and section in Figure 9.1(c).

9.2.1 Isolated Structure - Eland House

Eland House provides office space, varying from seven storeys on the west end to ten storeys on the east end, with basement and rooftop plant rooms (see Figure 9.1(c)). It is a reinforced concrete structure with solid flat slabs supported by circular columns. The building sits on a 1.5m thick raft, which in turn is founded on concrete piles 30-40m deep. Basement columns are rigidly connected to the raft, but stop short of the soffit of the ground floor structure, at which point the columns are enlarged to support the isolators and thereby the structure above (see Figure 9.2(b)).

The building is thereby isolated from the ground floor up, using TICO synthetic rubber bearings (type CV/CA), with a target rigid body natural frequency of 10Hz. The resilient bearing is made up from a laminate of plies of high tensile strength fabric bonded to plies of an elastomeric compound modified by the inclusion of cellular particles, built up in layers to a final production thickness of 25mm. The final bearing thickness is achieved in multiples of 25mm for the desired resilience, which in this case was to an overall thickness of 125mm. The building was completed in 1997.

Figure 9.1(b) shows the ground floor plan, and Figure 9.2(a) shows the basement plan indicating the alignment of the tunnel below. Plate 9.1 and Plate 9.2 show views of the isolated building.

9.2.2 Un-isolated Structure - Watney House

This is a 10-storey office building in reinforced concrete, with a basement and piled foundations. It was built circa 1960, and shortly after construction in the early 1960's, the foundations were modified to accommodate the then proposed alignment of the Victoria Line. Figure 9.3 shows a plan and section detailing these modifications, comprising piles for a new raft, and chemical consolidation of soils in the immediate area of the proposed tunnel where it passes close by the existing foundations of Watney House. Plate 9.3 and Plate 9.4 show views of this building.

9.3 Measurement Locations

Figure 9.1(b) and Figure 9.2(a) show the columns at which measurements were taken. Column P10 in Eland House was selected, along with both external columns of Watney House, closest to the Eland House site. Only measurements at column 'P10' of Eland House, and column 'A' of Watney House are presented, as they are both on the same side of the tunnel, at very similar distances from it (assuming the tunnel alignment given in the engineer's drawings are accurate).

Measurements include the vertical direction, as well as in two horizontal planes, referred to as tangential (T) and longitudinal (L), shown in Figure 9.2(a).

As the isolators for Eland House were placed on the top of enlarged basement columns immediately beneath ground floor, measurements were taken on the top of this basement column (see Figure 9.2(b)). Measurements were desirable on the raft, but this was inaccessible. Additional measurements were taken on the base of the isolated column at ground floor level (directly above the isolators), and close to the top of the column at ninth floor. Measurements on Watney House were only taken on the base of the columns at ground floor level. Whilst it would have been ideal to record the response of the top of the columns as well, access was limited.

The surveys were undertaken during 1997 when the building was completed, and fitted out. All measurements were taken late at night when the building was largely unoccupied.

It should be recognised that the horizontal distance between measurement points on isolated column 'P10' of Eland House and neighbouring un-isolated column 'A' of Watney House, of about 30m, causes the train vibration at the two points to arise from non-simultaneous events. For example, a peak in one time history may be due to a wheel traversing a defect in the rail, but the corresponding vibration level in the other time history will not necessarily be related to this event. This would present a shortcoming in any comparisons made, and represents a practical constraint, where the comparative levels were more likely to be train rather than system dependent.

9.4 Instrumentation

This consisted of four B and K 4378 accelerometers connected to B and K 2646 charge converters. These were connected to a Kemo signal conditioner, which in turn was fed to a National Instruments data acquisition card. Data acquisition was controlled and viewed using a program written by the Author in LabVIEW. Post-processing was undertaken using MATLAB®. The instrumentation and post-processing are described more fully in Chapter 6.

9.5 Vertical Measurements

Figure 9.4 shows a close up of time histories for a train pass-by. Simultaneous measurements are shown for the basement column, base and top of isolated column 'P10', along with measurements on the base of the un-isolated column 'A' at Watney House. The train vibration levels are seen to significantly exceed the background vibration shown on the right of Figure 9.4, except at the top of the isolated column. The increases above background levels, in spectral terms are shown in Figure 9.5, along with the instrument noise floor. Whilst the increase in level due to a train event is significant at all locations, it is limited up to 100Hz for measurements at the top of the isolated column, shown in Figure 9.5(a). The background levels are fairly close to the instrument noise floor, which implies that we must be cautious when interpreting any comparison of background levels. With this in mind, the time history of background levels on the right of Figure 9.4, indicates that the isolated building responds in a comparable way to the un-isolated building. Equipment with a lower noise floor as described in Chapter 6 would therefore have been desirable to determine the relative behaviour at such low levels.

The total and direct transmissibilities from the basement column to the base and top of the isolated column are shown in Figures 9.6 and 9.7 respectively. The building was targeted to have a rigid body vertical natural frequency of 10Hz. Yet the first peak in the transmissibility arises at 6Hz. This can be explained by examining the cross section shown in Figure 9.2(b). The basement column below the isolators is seen to rise some 5m above the raft, which is supported by piles. The axial stiffness of this basement column alone was determined using a finite element model (see Appendix 5.1 for model properties), such that when it acts in series with the stiffness of the isolators, it creates a natural frequency in theory at 7Hz, instead of the original rigid body natural frequency of 10Hz. It should come as a surprise to the designers, that the column below the isolators itself acts as a spring of comparable stiffness! A larger disparity between total and direct transmissibilities in the response of the top of the isolated column, compared to the base is seen which implied that the top was more influenced by uncorellated inputs from paths other than directly from the basement column below.

The total transmissibilities from the basement column to base and top of the isolated column are compared in Figure 9.8. At low frequencies the top of the isolated structure responds more strongly than the base, although shows more isolation at above 30Hz.

We can see how the total transmissibilities deduced from a train pass-by and background sample compare in Figure 9.9, from basement column to top of isolated column. Under background conditions the isolated structure largely shakes more than the basement column, although during a train pass-by, at low frequencies the shape of the total transmissibilities are largely the same, although clear differences arise at high frequencies. This implies that we must ensure that the vibration level from a train event must adequately exceed the background levels, otherwise we will see a transmissibility curve that is more indicative of background response. The similarities of response at low frequencies between train and background events implies that the structure is in a resonant condition in both cases, where magnification should be limited by damping to a similar extent.

The repeatability of total transmissibility from the basement column to the base and top of the isolated column, for 5 different train events is good, as shown in Figures 9.10 and 9.11. However, there is more scatter in the transmissibilities for different train events between the un-isolated structure and the isolated structure, shown in Figures 9.12 and 9.13. This is because of the distance between the two buildings, a practical constraint, which makes the transmissibilities more sensitive to variation in train events. A discussion of the mean values of these results is given in section 9.7.

9.6 Horizontal Measurements

Horizontal measurements were taken in two orthogonal directions, which lined up with the structural grid used at Eland House. These are labelled as tangential (T) and longitudinal (L) shown in Figure 9.2(a), although they do not align at right angles or parallel to the tunnel below, which is diagonal to the grid.

A close up of time histories for a train pass-by in the tangential axis is shown in Figure 9.14. It is clear that the isolated structure responds significantly less than the basement column and the neighbouring un-isolated column. The low background levels are shown on the right of Figure 9.14. The increase was found to be good at all locations except the top of the isolated column. Here instrument noise floor was found to be comparable to background levels, which would indicate instrumentation with a lower noise floor such as described in Chapter 6 would have been preferred. The background levels on the isolated structure are less than the neighbouring un-isolated structure in this measurement direction.

A comparison in the response of the top and base of the isolated column is shown in Figure 9.15, where the top indicates more attenuation at most frequencies. However, the results at the top of the column under a train event did not exceed background levels and was also close to the instrument noise floor, which indicates that attenuation to the top may even be greater than shown.

There is fair degree of repeatability of transmissibilities from the un-isolated column to the isolated structure shown in Figure 9.16 and Figure 9.17. The means of these results are discussed in the next section.

9.7 Comparison of Mean Results - Vertical and Horizontal

Five train pass-bys were analysed in the vertical and two orthogonal directions in the horizontal plane. This section describes the mean results, which are compared.

Figure 9.18 compares the total transmissibilities from the basement column to the base of the isolated column. In the horizontal axis attenuation reaches around 20dB in both cases, but arises at different frequencies. In the neighbourhood of 50Hz - 80Hz both directions show reduced attenuation, reaching some amplification in one of the horizontal directions. In the vertical axis there is clear amplification around 25Hz, reaching 10dB, with largely unity response up to 85Hz, followed by some high frequency attenuation. This is in distinct contrast to the results for the horizontal axes, which shows significant attenuation at low frequencies.

The mean transmissibilities from basement column to the top of the isolated column are shown in Figure 9.19. Deep attenuation is seen in both horizontal directions. In the vertical direction there is strong contrast in the behaviour at low frequencies. The vertical axis shows characteristic resonant modes starting at 6Hz, whereas the horizontal axes begins to show attenuation.

The mean total transmissibilities from the base of the un-isolated column of Watney House to the basement column beneath isolated Eland House are shown in Figure 9.20. The basement column shows distinct resonant modes at 20Hz and 40Hz in the "L" direction, although these are suppressed in the "T" direction. This difference can be explained by observing the site conditions. The basement column which extends from the raft, is in fact built into a blockwork wall, parallel to the "T" axis, which would suppress inverted pendulum modes in this direction. This can be seen from the basement plan in Figure 9.2(a). Whereas the "L" direction is at right angles to this wall, which is unable to restrain inverted pendulum motion for this direction. The low frequency resonant amplification in the horizontal direction clearly exceeds any in the vertical direction. The basement column beneath the isolated building responded at some frequencies 4dB more than the neighbouring un-isolated building in the vertical direction. This is likely to be due to the superstructure mass of the isolated building being 'dynamically detached' from the structure below, by the presence of the isolators. The horizontal measurements showed that the basement column below the isolators responded significantly (19dB) more than the base of the un-isolated column. This was attributed to inverted pendulum modes of the fixed-free basement column, which is again 'dynamically detached' from the superstructure owing to the isolators.

A significant comparison is made in Figure 9.21, which compares mean transmissibilities from the base of un-isolated Watney House, to the base of the isolated column. These generally show that in the horizontal plane the isolated building shakes up to 12dB less than the un-isolated building. Although there are areas of reduced attenuation, amplification is small, and mainly arises at higher frequencies. In the vertical axis there is significant amplification reaching 8dB with only some areas of isolation reaching 4dB - 8dB. This shows that the insertion loss for base isolation is better in the horizontal direction, with fewer and smaller disbenefits. The disbenefits in the vertical axis are serious enough to raise questions about the value of the small areas of improvement at specific frequencies.

Other than cursory inspections, detailed inspection were not undertaken to identify if any short circuits might explain the poor performance in the vertical direction. In fact a full inspection would also not have been possible, as the structural base isolation was designed in a way, that many of the necessary air gaps across the entire footprint could not be inspected, especially when the building was in its completed state. This in fact represents a shortcoming in the design of the overall base isolation system, as there are no practical means to check for short circuits which are a real risk in the construction of base isolated buildings.

9.8 Groundborne Noise

Internal noise measurements were only possible on the ground floor office of isolated Eland House, and could not be undertaken inside the adjoining un-isolated Watney House, due to access limitations.

However, in the ground floor offices above the tunnel, groundborne noise from underground train pass-bys was clearly detected in the low background levels that existed between 10pm - 12pm, at which time observations were made. This is evidenced from the sample measurements described in Table 9.1.

Table 9.1 Spot readings: Lmax (fast)

Train - felt and heard
Train - rattling window
Train - felt and heard
Train - just perceptible

The low frequency rumble from trains could be heard, but it was at a low level. It would clearly be unacceptable if this were a concert hall, and questionable in very high quality housing. Vibration on the mid-span of the ground floor was only just perceptible, though not unacceptably so for office or residential use.

9.9 Conclusions

This case study compared response of two similar isolated and un-isolated structures, which adjoin the same railway tunnel. The rigid body natural frequency of 10Hz was not clearly evidenced in measurements. In fact a peak in the response arose at 6Hz, and was due to the basement column below the isolators acting as a spring of comparable stiffness in series with the isolators. The basement column beneath the isolated building responded on the whole more than the neighbouring un-isolated building in both the vertical and horizontal direction. This implies that inertia and stiffness of the isolated building is uncoupled from the structure below in a frequency dependent way.

Response of isolated compared to the neighbouring un-isolated structure provides insertion loss. There were significant disbenefits due to base isolation in the vertical direction, and of the benefits seen, these were small, bearing in mind the difference that might be expected owing to inevitable dissimilarity in the two structures. There were no significant disbenefits in the horizontal direction, although the benefits, given the same dissimilarities, were much more evident than the vertical axis. This therefore implies that base isolation is at least effective in the horizontal axes, although the performance in the vertical direction, in this case, remains questionable.

Other than cursory inspections, no detailed examination was undertaken to identify if short circuits might be responsible for the poor performance in the vertical direction. Although, if short circuits are present, we might expect them to have some, although not equal effect on performance in the horizontal axes. A detailed inspection to validate the absence or presence of short circuits could not be undertaken with this building in its completed state. This has implications for quality control during construction, or the base isolation system be designed in a way that facilitates checking for short circuits across the footprint of the building upon completion.

Groundborne noise from trains was audible in the low background at night, within ground floor areas, although vibration was imperceptible. Base isolation at this site using TICO bearings would not present an adequate solution for very high quality housing or concert hall type use, although was suitable for the office use intended.