Surveying, Recording and Monitoring Historic Buildings
|Figure 1 Colour-mapped image of a point cloud of King Charles Tower on Chester City walls. (Image: Russell Geomatics/Donald Insall)|
The use of laser scanning techniques for surveying is now commonplace. Ten years ago a measured survey would have been carried out using a computerised EDM (electronic distance measurement) device such as a total station and ‘smart pole’, together with hand-held tools, but today’s surveyor is likely to be equipped with a tripod-mounted scanner.
Heritage professionals are likely to be more interested in having an accurate drawing on which to base their specification of works than in the process behind it. However, laser scanning is radically different from previous techniques, and it is well worth taking a moment to grasp the concepts, not only to take advantage of the new opportunities it offers, but also to avoid paying for unnecessary detail.
LASER SCANNING TECHNIQUES
The technique requires a scanning head to be mounted on a surveyor’s tripod. The scanner spins at very high speed while a low-energy laser fires a reflecting beam with extraordinary precision, recording up to 1 million points a second. The density of these is adjusted according to the purpose of the survey, with a typical spacing of 1-3mm. There might, for example, be 20 million points in a survey of the front facade of a modest Grade II listed cottage. The tolerance of the position of each point is typically 1-2mm.
Together they effectively describe the surface and are known as the point cloud. Figure 1 shows a colour-mapped image of the point cloud for King Charles Tower on Chester’s city walls. Safety is a common concern for neighbours and bystanders. Lasers used for such work are of Class 3R or lower intensity in accordance with IEC standard 60625-1 and under normal use the beam is not harmful to the human eye. Legislation may require warning notices to be displayed while site work is carried out and the surveyor should have a method of work that mitigates exposure.
|Figure 2 Typical survey equipment showing spheres mounted on tripods to correlate datasets taken from different locations (from either side of the river, for example)|
Obviously, the device does not have X-ray vision. To give a complete picture it must have sight of the features to be surveyed (as with the more traditional total station), so a number of different set-up positions will need to be adopted inside and outside the building. It may need to include roof voids and positions on or overlooking the roof itself if an accurate roof plan is required. Laser survey equipment has become lighter and can now be mounted on an extendable pole, although sway may result in error. Even so, it is often impossible to avoid areas of shadow on the point cloud where surfaces are hidden behind other fabric, and data is lost. In such cases some assumptions must be made later to fill in the gaps.
The set-up position does not need to be located over a survey ‘station’ (a nail head or pin), which is traditionally used to tie the survey together. Instead, the survey company will commonly use their own objects, typically spheres, to correlate the dataset from one location with that taken from another, as shown in Figure 2 (above right). Each piece of survey is then fitted together like a jigsaw so that the edges match to form a whole.
The scanning of motorways and railways is now undertaken from moving platforms, vehicles and even aircraft but, due to the lower tolerance on such data, this method is unsuitable for historic fabric.
CURRENT SURVEYING PRACTICE
Survey companies have moved rapidly to embrace laser techniques because they reduce both the costs and the risks associated with site work. The time needed to record data can be as little as ten minutes per location. By reducing site activities and transferring them to the office, the influence of unpredictable factors such as bad weather are mitigated.
The benefits are significant when considering large building volumes and spaces where detail at height is important, such as cathedrals, tall facades and historic civil engineering structures. There is usually no need to gain access at height in order to register their dimensions. The precise shape of a historic vault, a bulge in a wall or the irregular spacing of timbers across a ceiling can be measured from ground level.
Laser scanning therefore provides a new approach for the historic building specialist and a new way of visualising and exploring historic fabric. Its principal advantages over previous methods include:
- Recording detail from a distance The shape and condition of decorative stonework, corbels, lintels and other features can be reproduced by the surveyor at large scale (1:5 or even 1:2 if necessary) with a tolerance equal to or better than that obtained by close physical measurement.
- Inspection in low light The process is not dependent on the human eye so can be carried out at night when a building or site is unoccupied, or with very low levels of internal light.
- Access to a complete computerised record Conservation professionals have access to all the gathered data in scalable form on their own computer. They can jump from one survey position to another in a 3D environment, interrogating floor levels, lintel heights and other dimensions.
CHOICE OF SURVEYOR
Different surveyors specialise in different scales and types of work, so knowledge of the required outputs and attention to historic detail is essential. A mixed portfolio that includes small and medium sized projects and strong experience of historic fabric is usually a good clue to a surveyor’s suitability.
SPECIFICATION OF OUTPUT
Some time-consuming processing occurs after a laser survey is undertaken. It is therefore more important than with older survey methods for the professional to specify exactly what is required. A specification for conventional output of 2D drawings should include:
- a description of the purpose of the survey
- the physical extent of the work, including roofs and voids
- the point density and point tolerance
- the 2D drawing series of plans, elevations and details, if necessary using photographs or pre-existing survey records to help clarify the work
- parameters that describe the tolerance of detail on the drawings.
The tolerance of the detail can be described, for example, by the required scale of drawing (1:100 up to perhaps 1:5). Quoting this will help the surveyor to decide how much detail to include. Further guidance is given in sections 5 and 7 of English Heritage’s Metric Survey Specifications for Cultural Heritage (see Recommended Reading).
It is important to define what project-specific detailed drawings may be required at the beginning of the survey, so that the surveyor can adjust the density of data collected and set-up points to focus on specific needs.
FROM LASER SCAN TO DRAWINGS
The scan collects large volumes of data which are stored in compressed format on the device’s hard drive. The data is then downloaded and processed to become the point cloud in a process known as registration, undertaken using software such as Cyclone™. During this process, spurious points are removed and the point data is converted to a standard transfer format. The point cloud might typically contain between one and ten billion points that describe the building surfaces inside and out.
|Figure 3 Detail of stonework drawn at 1:50 and enlarged to show discrete linework (Image: Greenhatch Group)|
The process of reducing this to 2D drawings or 3D models usually involves thinning this to a lower density. In the case of 2D drafting, a cutting plane is defined and the data exported to form the drawing using proprietary software such as CloudWorx™. The process of creating the 2D image, known as vectorisation, is a simple but rather laborious process of join-the-dots. An enlarged detail is shown in Figure 3 (left).
Clearly, the greater the accuracy required, the less thinning-out is undertaken and the more dots there are to join. So it is essential that the surveyor knows what resolution is required from the start. Quoting a drawing scale is still a good way of expressing this, even though CAD effectively functions at 1:1.
Scanning is particularly suited to recording highly irregular surfaces such as timber frames and medieval stonework, and their individual components can be clearly identified from the scan. However, it is often necessary to use photogrammetry in conjunction with laser scanning to trace more uniform areas of brick, terracotta or stonework. If this level of detail is required, for specifying repairs to individual stones for example, then this requirement should be stated at the outset for including in the pricing. The use should also be discussed in detail with the surveyor prior to site work.
OWNERSHIP AND TRANSFER OF DATA
Normal principles of ownership and intellectual property usually apply to the output. The survey company retains this and the purchaser is typically given a royalty-free licence to use it for the purposes defined.
3D point cloud data is usually available to the purchaser if requested. However, the amount of data is large, sometimes running into terabytes, so an external hard drive is normally used to transfer it.
A SCALABLE IMAGE AND A 3D PHOTOGRAPH
There is nothing quite like having photographs to record and look back on a site visit, or explain the project. If photographs were scalable, we would be able to confirm dimensions and take levels that perhaps are not covered on our 2D drawings, however well specified.
The program TruView™ is a very powerful tool that does just that. Truview takes the scan data and effectively produces an image of the building on an office PC. This can be rotated, enlarged and interrogated for dimensions and levels. It is a very useful application that provides good visualisation. Small architectural details can be enlarged for closer inspection. Structural defects such as cracks down to perhaps 1-2mm can be viewed, and sagging in beams and the bulge in walls can be read, even though the professional may be unable to see them on site.
The point cloud is not the only site data that can be recorded while on site. A 360° colour photographic image is often specified in conjunction with the scan. These photographs are not currently scalable, but by combining the digital photo with the scan image so that both are registered by the computer program, the photo itself appears to be scalable. Scanning techniques are being developed to record the colour directly using the spinning laser, before registration.
|Figure 4 Point cloud (above) and surface model (below) of an existing school facade|
The cost of on-site laser scanning is now essentially hidden in the survey itself, so there is no cost premium. The whole process of 2D drawing production is no more expensive than by traditional total station techniques, assuming that normal drafting tolerances are specified.
The production of drawn output from the collected point cloud data will normally represent about 60-70 per cent of the cost of the overall survey, but this may rise to 90 per cent for detailed stonework, showing just how important it is to define the survey parameters.
Some consultants are tempted to produce the drawn output themselves from the point cloud. This can achieve good results but may prove more expensive. Direct use of the cloud by the consultant is more normally associated with surface and solid modelling.
Photographic colour imaging currently requires different equipment to be mounted to the tripod and so roughly doubles the time spent at each site set-up. It also adds to the registration process and will add 10-25 per cent to the overall cost of a survey.
3D modelling can be economical for basically rectilinear and recent fabric. The simple facade in Figure 4 (above) was scanned and modelled, out-of-plumb and complete with bulges, for £1,350, ready for incorporation into the BIM model.
DRAFTING INTO 3D
The use of Building Information Modelling or Management (BIM) has been declared as a government objective in the procurement of design. Although capable of much greater sophistication, reduced to a minimum, BIM is effectively 3D drafting. When existing building fabric is to be repaired or modified, laser scanning provides the key by which survey data is pivoted into the model. By surface modelling from the point cloud, a historic building can reappear in reduced, filtered, and rendered form in the 3D drafting model, yielding very considerable benefits in terms of fit and visualisation.
|Figure 5 Point cloud (right) and surface model (left) of the hull of the Tudor warship Mary Rose|
This process requires experience and it is best to start simply. A number of survey companies will build a surface model from the point cloud, in the same way that they produce 2D drawings. The tolerances on a 3D surface model need to be carefully defined if the model is to be reliable and the deviations should be clearly understood (for example, permitting a 5mm or 10mm maximum deviation of the surface from the cloud).
The modelling process is much more expensive than creating 2D drawings. It may take a number of weeks to produce a model of a complex building or a structure like the Mary Rose (see Figure 5, left). Even so, some spaces are likely to escape survey and so cannot have surfaces fitted to them.
Drafting software such as Revit™ can now accept point clouds directly into the 3D model. The manipulation of the cloud by drafting platforms looks set to develop rapidly over the next few years, as it becomes easier to use scanned survey data for existing and historic buildings on office PCs.
RECORDING HISTORIC FABRIC
The use of laser scanning for recording and archiving is now well established and English Heritage (EH) has produced good guidelines for the professional (see Recommended Reading). Concern has been expressed in the past over such methods because of the stability of the electronic archive, but this issue has been largely addressed by the National Monuments Record. EH has recently commissioned a comprehensive laser scan of Stonehenge and work at Ironbridge is due to commence shortly.
The choice of format in which the archived data is kept remains an issue because different manufacturers have different formats. The common choice remains the rather inefficient ASCII format which generates very large files for storage. The ASTM E57 format currently in development may provide an effective alternative.
|Figure 6 The final product: a working drawing showing a cross section of an archway through Chester city walls. Accurate, detailed drawings like this can be used for a wide variety of practical applications, from preparing specifications to monitoring and recording. (Image: Russell Geomatics/Donald Insall)|
Scanning provides a highly accurate contoured surface of buildings and structures. By repeating a scan a few months or perhaps a year later and overlaying one scan over the other, three dimensional movement in the surfaces can be detected. This provides a powerful tool for monitoring structural behaviour, given that it can be achieved without physical access to the walls. Contours (or ‘isopachytes’) can be produced using a program such as Geomagic™ that shows the difference in movement.
This technique can be used at the sub-millimetre level to record the decay in surfaces such as brickwork or even decay in objects, using a more sensitive group of laser scanners that operate on the principle of triangulation, sited perhaps one metre from the object. A tolerance of 0.5mm is currently achievable.
- D Andrew (ed), Metric Survey Specifications for Cultural Heritage, English Heritage, Swindon, 2009
- D Jones (ed), 3D Laser Scanning for Heritage, English Heritage, Swindon, 2007
- RICS, An Introduction to Terrestrial Laser Scanning, London, 2006
The author would like to thank colleague Daniel Niziolek, Andrew Dodson of Greenhatch and Paul Bryan of English Heritage for their assistance in the preparation of this article.