Industrial Sponsorship

Dynamic Performance Investigation of Base Isolated Structures

By Ather K. Sharif

11.1 Conclusions

Whilst many buildings have been base isolated to date, since the original concept in 1916, surprisingly little research has been done to evaluate or optimise performance.

This thesis examined dynamic performance and wider issues of Base Isolation as a foundation to drafting a new International Standard on the subject. Dynamic performance issues were addressed using case studies. Broader issues were identified from an understanding of factors relating to source, propagation path and receiver.

Source and propagation characteristics of vibration from railways are well understood, although theoretical models rarely treat the entire source to receiver path, and there remain concerns over the actual accuracy of both empirical and theoretical models.

Vibration from railways is unlikely to damage buildings, although human perception of it will continue to raise concerns, particularly from freight traffic. Base Isolation is therefore unlikely to be used to address the direct effect of vibration on the structure.

Perception of vibration and groundborne noise is likely to be the main reasons to consider Base Isolation. Evaluation of Vibration Dose Values in accordance with BS6472 (1992) is unlikely to ever justify a need to Base Isolate a structure. This was demonstrated using measurements on railway platforms, which showed that most sites adjoining a railway would pass the standard and yield a low probability of adverse comment. Yet it is known that people do complain of vibration at lineside properties, and this discrepancy was used to justify a re-examination of the guidance in BS6472.

Certain equipment and processes are very sensitive to vibration, and may well be a reason to consider Base Isolation. However, computers are unlikely to be adversely affected by typical vibration levels at lineside, despite conservative limits put forward.

There are a variety of vibration control measures that should be considered as either alternatives to Base Isolation or as supplementary measures. Limiting the amplification at the mid-span of floors is likely to be the most worthwhile aim in a design.

Theoretical comparisons show that SDOF models should not be used to assess the behaviour of base isolated structures, where MDOF models are essential. A dynamic vibration absorber system was shown to be potentially relevant in respect of floor response in Base Isolated structures. Transients may be dominant in a waveform (e.g. close to jointed track). Transient response is different to steady state, and requires appropriate analysis. Effects of starting transients and time to build up resonant response will be important, particularly in low damping, low natural frequency systems (e.g. low frequency Base Isolation systems using coil springs and in long span floors).

Important considerations in selection of instrumentation and digital data acquisition have been shown necessary to provide quality data. Trade off between statistical accuracy and frequency resolution was shown to be acute for the discrete event of a train pass-by, and in reality such events are non-stationary (an analysis constraint).

The first case study (Chapter 7) determined insertion loss due to base isolation as the difference in response between un-isolated and isolated states. Alternative isolators, of increasing stiffness and damping (GERB steel coil springs @ 4.5Hz, BTR laminated natural rubber bearings @ 8Hz and TICO synthetic rubber bearings @ 12Hz) were evaluated, giving reducing insertion loss. It was shown that base isolation could yield a worthwhile reduction in vibration and groundborne noise, although resonance characteristics of the supported structure or poor choice of isolator can detract from this. Simple measurements across the isolators, which are often the only measurements possible after a base isolated building is constructed, was in this case shown to be a reasonable indicator of insertion loss. An error arises due to a difference in foundation response according to how the building mass is coupled (isolated or un-isolated). Even using the softer GERB isolators, groundborne noise could still be heard in a low background. Base Isolation may not always solve the most demanding applications.

The second case study (Chapter 8), a complex of buildings, compared the response of isolated (BTR @ 7.5Hz & 9.5Hz) and un-isolated buildings that were otherwise similar or identical. Short circuits were discovered in some isolated columns, reducing performance by a small amount in comparison with neighbouring columns free of any short circuits. The small difference was surprising and led to a suggestion that the grout might have cracked due to differential movements. The inertance of isolated columns was shown to be greater than an otherwise identical but un-isolated column, implying that an isolated column is easier to excite, and was borne out by comparisons of the response due to services plant. Modal damping was found to be broadly comparable between isolated and un-isolated columns, that were otherwise identical, and was of a hysteretic type. Point inertance of pile caps below an isolated column were shown to be greater than below an un-isolated column. Correlation with theory showed that the pile cap below an isolated column responded as though the inertia of the column was not there (for a limited frequency range), but the pile cap below an un-isolated building was influenced by the inertia of the column above. Column models with lump masses to represent floor loading were found to reproduce the low frequency characteristics, and were a significant improvement over SDOF or uniform bar models, although complex models would be needed to explain the larger part of the spectrum. Base isolation was not considered to be worthwhile in dealing with the predominately low frequency ground vibration that arose at this site from the railway at grade.

The third case study (Chapter 9) compared the response of isolated (TICO @ 10Hz) and un-isolated buildings that were otherwise broadly similar, over an underground railway. 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 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 implied that base isolation was at least effective in the horizontal axes, although performance in the vertical direction remained questionable. The building as completed did not facilitate access to check that there were no short circuits. This implies that there should either be strict quality control during construction to avoid short circuits or structure detailed in a way that permits subsequent site inspection to verify that there are no short circuits.

Clearly lack of adverse comments should not be used to infer good Base Isolation performance. According to the objective of this thesis, a draft International Standard has been prepared and submitted to ISO Committee (TC108/SC2/WG3) for consideration.

11.2 Recommendations for Further Study

Reducing contact area of a structure with the ground can reduce groundborne noise. It would be interesting to examine whether this approach alone could be used as an economical means of groundborne noise control for some situations.

Point inertance of a variety of sub-foundations, with buildings that are traditionally coupled needs to be further compared with buildings that are isolator coupled.

There is a need to develop theoretical models that can examine the response of foundations to an incoming wave-train, over a broad frequency range. The analysis should then be developed to examine the interaction of the supported structure (both traditionally and isolator coupled), considering the correlation of inputs. Laboratory scaled models would be an appropriate means of investigation.

Further research on modal damping of isolated and un-isolated structures would be desirable and appropriate damping models for a structure remains a priority.

The trade off between increasing stiffness to move natural frequencies of walls and floors above dominant spectra in the ground vibration spectrum needs to be weighed up against the potential effect that groundborne noise may be increased.

More research needs to be directed on material properties for the soil and rubber bearings at very low strains, comparable to the magnitude expected at lineside, recognising that high frequencies are associated with significantly smaller strains.

A simple, electrical resistance ‘type’ test, for quickly identifying the presence of short circuits in a Base Isolated structure would be desirable.

Investigation into adverse effects of short circuits needs study. Wind sway restraint devices should be evaluated to ensure that they do not compromise performance.

Despite this knowledge, a most recent Hotel development in London over underground railways was isolated using TICO bearings with a high target natural frequency of 12Hz.

Since the original concept of Base Isolation, there has been a trend to use lower rigid body mount frequencies in an expectation of improved performance.

Yet given the large number of buildings that have been Base Isolated in the UK and around the world, surprisingly very little serious investigation has been undertaken into the dynamic performance of Base Isolation systems. Most of the papers on the subject refer to the use of SDOF models, and quote high levels of expected performance, but which are rarely validated by site measurements.

This thesis has examined the dynamic performance and wider issues of Base Isolation as a foundation to the drafting of a new International Standard on the subject. The dynamic performance issues were addressed using case studies. The broader issues were identified from a complete understanding of the factors relating to the source, propagation path and receiver.

The first case study described an experimental investigation into the performance of Base Isolated structures above an underground tunnel using a test room, initially un-isolated, and then isolated onto different resilient elements using GERB coil springs, BTR steel laminated natural rubber bearings and then TICO synthetic rubber bearings. These tests for the first time provide the actual insertion loss, demonstrating that Base Isolation was effective in improving the situation over that in its un-isolated condition. The tests showed that a lower Base Isolation natural frequency did significantly improve performance, particularly in terms of perceptible vibration although the difference between different isolators was less significant in terms of groundborne noise. The tests highlighted that a poor choice of Base Isolation natural frequency, such as in the case using TICO bearings could significantly reduce the value of Base Isolation at certain frequencies. The resonance modes of the structure clearly reduced the performance and significantly deviated from predictions using SDOF models which are frequently used in Industry. MDOF models were however shown to adequately represent the key features in the response and therefore represent a practical tool for analysing the behaviour of Base Isolated structures.

The experiments showed that a disbenefit of Base Isolation arose at specific frequencies, due to the restraining effect that a building mass has on the sub-foundation when using traditional foundations, which is lost when the building is Base Isolated.

Uniquely, the entire final building was also economically tested in its un-isolated and isolated conditions, by adequately pre-stressing the coil springs to use the stored energy to raise the building. These tests provided the actual insertion loss for Base Isolation, showing a significant and worthwhile improvement reaching 30 dB in terms of vibration and groundborne noise at some frequencies. The overall A-weighted sound pressure levels were reduced by a mean value of 18 dBA, representing a significant and worthwhile improvement.

Impact testing was able to reproduce the response characteristics of the Base Isolation system, although its wider use on a larger structure has been addressed separately.

These tests proved that whereas a site would have been uninhabitable for residential occupation and even office use in its un-isolated condition, it was turned into a high quality environment when isolated. Even so, the resulting environment would not have met the standards required for a concert hall, and therefore shows that there are limits to the use of Base Isolation alone as a solution to cater for all sites and uses.

The second case study examined the dynamic performance of large concrete framed buildings adjoining a railway at grade, in which two of the blocks were Base Isolated using BTR steel laminated natural rubber bearings, for a natural frequency of 7.5 Hz and 9.5 Hz.

A Base Isolated building was shown to respond more than a neighbouring un-isolated building under ambient conditions due to its mechanical services. This had a further significant implication, in that the performance of the Base Isolation system due to the external source would be underestimated unless the signal of the external event significantly exceeded background levels. This ideal condition could not be achieved for all the train events, and was dependent upon frequency.

The transmissibilities from the pile cap to the base and top of the 30m tall column, showed that a fixed-free mode of the column existed with a natural frequency close to the original rigid body mount frequency selected for Base Isolation. A simple bar model that is sometimes used to model a column would not replicate this characteristic, but a model which included the storey loading was shown to provide significantly improved correspondence with site measurements.

The deep attenuation predicted between resonance modes of an individual column could not be realised in practice, because the large structure demonstrated so many more resonant modes. Modal density was therefore introduced as highly relevant parameter. This would be greater for a larger structure, implying that Base Isolation may prove to be more beneficial in a smaller structure in which the modal density within the dominant frequency range of the source is smaller. That is, there would be fewer resonant response opportunities that could compromise the effectiveness of Base Isolation in a smaller structure.

Impact and eccentric mass shaker testing was used to demonstrate that an isolated column was easier to excite than an un-isolated column, which confirmed that building services within an isolated building would cause a greater response, which was borne out by the ambient measurements.

The performance of the Base Isolation system could not be simply based upon comparisons with measurements on the pile cap below the isolators. This is because the pile cap below an isolated column was shown to respond more than the pile cap below an un-isolated column. This was uniquely confirmed using 'point' inertance results with an eccentric mass shaker.

The measured 'point' inertance compared well to the predictions using a theoretical model, and showed that the pile cap below the isolated column behaved as though it were unloaded. Whereas the pile cap below the un-isolated column behaved more as though it were loaded. This therefore clearly indicates that the isolators dynamically detach the supported mass from interacting with the pile cap in a frequency dependent way. The theoretical analysis highlighted that some pile caps, which are often at face value treated as rigid, should in fact be treated as flexible.

Impact or shaker testing could not be used to simulate the effect of a rail source in a large structure, such as this framed building. This is because the input that arises local to one pile cap undergoes three dimensional decay into the structure. Whereas a rail source will excite a number of pile caps, and whilst each column would shed energy to neighbouring columns, it will also receive energy back from others.

The building with the lower Base Isolation natural frequency performed slightly better. However it was concluded that Base Isolation in this case study was not very effective, and whilst there were some limited benefits at some frequencies, there were many more disbenefits.

Base Isolation using a natural frequency of 7.5 Hz and 9.5 Hz was therefore not considered appropriate, due to the multitude of resonant modes of the large framed structure which fall within the predominately low frequency ground vibration spectra that arose at this site. However the vibration on the midspan of floors only borders just perceptible and there have been no complaints in the 10 years of occupation. Clearly, lack of complaints or a low level of vibration should not be used to infer performance.

Small localised areas of short circuits were discovered highlighting a practical risk in the design and construction of Base Isolated buildings.

The third case study compared the response of two similar neighbouring buildings over an underground railway, but with one isolated using TICO bearings with a rigid body natural frequency of 10 Hz.

In the vertical direction, the base of the isolated building showed disbenefits reaching 8 dB, and at some frequencies an improvement reaching 4 - 8 dB compared to the neighbouring un-isolated building. However, in the horizontal axes, given the same inevitable dissimilarities in the structures, the isolated building showed no disbenefits, but a significant improvement reaching 12 dB.

The column below the isolators was shown to respond more than the base of the un-isolated building, indicating that the isolators detach the constraint imposed by the superstructure. The base of the isolated building responded 10 dB more than the un-isolated building under ambient conditions in both the vertical and horizontal axes. Although the services in the two buildings are not identical, it does imply that an isolated building responds more due to internal sources of vibration.

It is clear that Base Isolation in this case was significantly effective in the horizontal axes, and this would have had a benefit in reducing groundborne noise. The performance in the vertical axis was however more questionable.  

This research has shown that Base Isolation can achieve very significant and worthwhile improvements in terms of vibration and groundborne noise. However there are cases where Base Isolation can be inappropriate, and where performance falls seriously short of the expectations in Industry.

The benefits and disbenefits of Base Isolation should be assessed in relation to what might have been achieved had traditional foundations been used. The insertion loss that is derived in this way is not an absolute measure of the performance of Base Isolation that could be applied elsewhere. This is because it will be related to a site specific reference situation, and would vary according to the type of traditional foundation that was assumed.

Measurements across the isolators are often the only practical measurements that can be taken to assess Base Isolation performance. This research has shown that in some situations (case study 1), it can be taken as a reasonable indicator of insertion loss for Base Isolation. The errors, which are frequency dependent, arise due to the differences in mass effect on the sub-foundation, depending upon the way in which the superstructure is supported (isolated or un-isolated). In some circumstances these errors may be large (case study 2), and therefore depend upon the type of superstructure, the isolators and the sub-foundation.

It is important that Industry recognises that Base Isolation is not a conservative solution, and the potential disbenefits need to be weighed up against the potential benefits, with due regard to cost effectiveness along with proper consideration of alternative options for vibration control.

A case for a new standard has been made by the Author, to specifically address the wider issues concerning Base Isolation and in particular the dynamic performance.

As a conclusion to this research, a first draft International Standard was prepared and has been accepted by ISO/Technical Committee 108/SubCommittee2/Working Group 3 into the process for standardisation.  

Short circuits remain a practical risk in the construction of base isolated buildings.

Clearly lack of adverse comments in the 10 years of occupation should not infer good performance, as is sadly the acid test used in Industry to date.

This research uses case studies to highlight key features in the dynamic behaviour of Base Isolated structures. The building responses are compared with the behaviour predicted by SDOF models, which have and largely continue to be used. Finite element (MDOF) models are used to show that these provide significantly improved correlation with site behaviour. Impact, shaker and ambient testing have also been used to establish some of the dynamic characteristics of a Base Isolated structure.

Chapter 2 describes the mechanism by which ground vibration is induced by the passage of trains, the propagation characteristics and means of attenuation in the propagation path. It describes the alternative prediction models in use.

Chapter 3 describes the effect of vibration on the building structure, the effect of vibration and groundborne noise on the occupants and any sensitive equipment or process. It reviews the use of Vibration Dose Values described in BS6472 (1992) as a means of evaluating Human response and makes a case for a revision of the standard.

Chapter 4 describes alternative vibration control measures that can be considered as either an alternative to Base Isolation or as supplementary measures.

Chapter 5 describes some theoretical aspects of isolation, describing the benefits of MDOF over SDOF models, the relevance of a Dynamic Vibration Absorber system, the effect of transient pulses, consideration of starting transients/steady state response and the relevance of the finite time to build up resonance.

Chapter 6 describes considerations in the selection of appropriate instrumentation for the case studies. It describes the method of processing used to improve statistical confidence in the spectral estimates, and the way that transmissibilities can used to evaluate performance of Base Isolated structures.

Chapter 7 is a case study of buildings over an underground railway, which describes a unique experimental investigation into the performance of Base Isolated structures, using a test room to evaluate the insertion loss for alternative isolators with different natural frequencies. For the first time it provides the actual insertion loss of a Base Isolated building, in terms of vibration and groundborne noise by comparing measurements in its un-isolated and isolated state.

Chapter 8 describes a case study, comparing the performance of both similar and identical neighbouring isolated and un-isolated buildings adjoining a railway at-grade, where Base Isolation utilises BTR natural rubber bearings. It shows how a simple bar model to describe the axial modes of a column can be improved and correlated better to site measurements when the effect of storey loading is modelled. It evaluates the method of impact and shaker testing to determine the dynamic characteristics of a Base Isolated structure, and uniquely provides the point inertance of a pile cap below an un-isolated and isolated column which are compared with theoretical predictions.

Chapter 9 describes a case study of a building over an underground railway, which is Base Isolated using TICO synthetic rubber bearings. It compares the performance of this isolated building with that of a broadly similar but un-isolated building. Whilst there were differences between the two buildings, the insertion loss comparison between the vertical and horizontal direction could be compared under the same circumstances.

Chapter 10 describes the standards process, the scope and limitation of an existing British Standard (BS6177, 1982), and identifies the need for a new standard to deal with both the wider issues of Base Isolation but also the important aspect of performance. It concludes with a first draft proposal for an International Standard, which can form the basis of a contractual agreement between the Designer of a Base Isolation system and a Developer, to ensure that the design decisions are accountable.

Chapter 11 refers to the conclusions drawn from each chapter and gives recommendations for further work.