A brief guide to modern measured building surveys.​

Today’s urban spaces are complicated beasts. If you’re not tussling with the physical space itself to accommodate all your usage requirements, you have planning, building regs and environmental considerations to face. The ever-increasing density of our developments mean a more critical focus is needed to ensure the size, shape and aspect of our internal spaces meets the needs of the inhabitants while respecting our neighbours – near and far – and their legitimate claims to the amenities enjoyed in their own spaces. The forgiving element of ‘space’ is mostly a long-forgotten luxury within our cityscapes. Modern, innovative design within tighter, multi-purpose spatial arrangements has bricked Architects into a tight corner of fractional tolerances. Where every millimetre counts, accuracy is key.

While CAD systems have done much to tame inaccuracies of design in the past 40 years, the survey data available to facilitate and justify tight design tolerances has similarly developed. The Architecture / Surveying symbiosis is in constant evolution, with advances in one justifying advances in the other. However, despite the huge technological advances we have already experienced to enable millimetric precision, the process continues…

The evidenced story of surveying starts around 2700BC in Egypt. The Pyramid of Kufu was built so accurately square and oriented to the points of the compass that it is unlikely to be coincidence. Tomb tablets found, dating from around 1400BC, depict land measurement, city construction plans and chain-men measuring a grainfield using what appears to be a rope with knots or marks at uniform spacings. The Romans, taking Egyptian instruments and improving on them, developed the ‘Groma‘ which was a tool for measuring right-angles and sighting straight lines. In the centuries since, instrumentation has come a long way but the basic principals have barely changed. The word ‘theodolite’, first recorded in 1571 by English mathematician, Leonard Digges, comes from the Greek ‘to look attentively upon’. And we’re still attentively looking…

Jumping forward somewhat in my walk through modern surveying, I’ll recommence where I started my own journey – in the 1980s. This was a time when CAD was developing at pace and being adopted in drawing offices worldwide. EDM (electronic distance measurement) was becoming more affordable and portable and being integrated into survey instrumentation to create ‘total-station theodolites’, as well as being deployed in handheld units. The new breed of instruments didn’t, necessarily, improve the ‘building control’ accuracy achievable by a skilled surveyor, but it did open the field to less-skilled (or less mathematically inclined) participants and reduced survey times dramatically. A total-station instrument traverse around (and/or through) a building could express building control points accurately, reliably and quickly. The output was a series of X, Y & Z coordinates – points on a CAD screen – defining building corners, openings and any other measurable feature. It was this ‘control’ that was the bedrock of all surveys. And still is to a large extent. Basic processes and accuracies of total-station instruments have marginally improved over the years – the main advances in operational terms being seen in user interfaces, portability and compatibility.

Traditional and modern site measurement surveys
So, how do we create a survey from dots on a screen? We add in hands-on sketch-and-measure – a trusty A3 sketch-board, a 2m wooden rule and, if you were pushing the boat out in the early years, a handheld EDM. A retracting tape-measure could be employed, but the one-handed operation possible with the wooden rule was a time-saving and practical preference for many. Sketching skills were a benefit in order to depict the building on paper as projected to be drawn in CAD. The production of these on-site approximated representations of building lines on paper required a competence gained with experience. What do you show? How do you represent intricate features efficiently? And, importantly, in a balance of client requirements and expedition, what do you not show? It was the adopted norm to spend as much time sketching as measuring to ensure a sufficiently detailed, coherent 2D model on paper. Every nook and cranny that needed to be shown needed to be measured and check-dimensioned so there was always a backup. Wall-door-wall… overall dimension. Wall-window-wall… overall dimension. EDM tie-diagonals, window cill and head heights, ceiling heights, services, sockets, pipe-runs, etc, etc… Typically, features were measured to the nearest 5mm and one would aspire to complete up to 5 fully-detailed, full page A3 sheets in a day. It was a surprisingly physical operation and a skill to behold when seen done efficiently.

Back in the office, we have a processed spattering of dots on our screen and a ream of neatly-dimensioned sketches next to us. Where to start our survey drawing? With your longest, single straight line, taken as a best-fit between a series of points… at each floor. And then checked against other, notionally perpendicular or parallel lines. Lots of lines, lots of angles. And the likelihood of all of these disparate lines on your screen being at a neat, single angle are minimal, so there is a game to play with homogenising angles, discarding lines, splitting differences, eliminating possible erroneous points and sometimes guesswork to see if it fits, drawing/deleting/redrawing. It’s a battle to marry the points on the screen with the measured dimensions – like trying to persuade same-poled magnets together- always with a tolerance because it’s never going to fit exactly. Within 2cm was OK (and was the grudgingly-accepted tolerance). Within 1cm was comparative perfection. On a large building a whole day could be spent just playing with these lines to arrive at a neat and coherent set of external walls as the basis for the real work to come – the internals.

If the luxury of internal instrument control was provided, that made subsequent survey drawing levels-of-magnitude easier. If the site didn’t lend itself to internal control, or the occupants simply didn’t allow, then the work was all yours to do in the office. Drawing surveys accurately from sketches and external-only control is not something that can be replicated with any degree of accuracy by the un-practiced. To achieve the expected accuracies takes experience, time, targeted trial-and-error, and a 3-dimensional vision that cannot be taught. With so many unknowns in position and angle, achieving correct positioning of walls and features across a building requires that linework trials be undertaken, working on angular assumptions initially, then honing positions and angles until all dimensions in both horizontal planes work within tolerance. This takes CAD expertise to be efficient, and knowledge of the building style, age, construction and context all helps, combined with interrogation of information from as many sources as you can find – photos, what happens on floors above and below, memory of being on site, notes on sketches, indicators from external lines, previous similar projects, intuition, more trial-and-error…  In all, internal survey drawing of a building is a 3D puzzle where any move has a knock-on effect and (tolerance notwithstanding) only one correct answer.

Handheld electronic survey systems were being developed in the ‘90s which pieced together dimensions in real-time on site. These systems certainly serve a purpose and have their valid applications. The main benefit being the time-saving aspect of cutting out the sketching and reducing a hefty amount of office time. However, in saving time on omitting the human processes, the very thing that the experienced human surveyor brings to the table that is of value in terms of accuracy, detail and coherence are also omitted. Fully-weighted, informed decisions on the push-and-pull between angles and dimensions, taken with reference to a plethora of contextual data, is still very much in the human realm. Accuracy, on these terms, cannot be automated.

However, time and technology has moved on. Arguments about the efficacy of the above measurement and drafting methods are made moot by the advent of the laser scanner. From inception in the 1960s to its continuing evolution today, this exciting technology is now considered the default data-capture method for commercial measured building surveying. In the past 20 years, the laser scanner has seen a fall in cost and consequent adoption by anyone serious about accurate ‘measured’ surveys. Although existing in different formats, the typical survey scanner, roughly the size of a football, sits atop a tripod and can capture a million points per second, bounced off solid objects in a sphere around itself (with a small blind-spot under its feet). Each point is defined by relative X, Y & Z coordinates as well as (among other things) colour of the measured surface. The point-cloud produced models the whole space in 3 dimensions, with data detailed enough to read the words in a book. Compared to traditional data-capture methods, the information density is profoundly greater and accuracy much more reliable. Site time, while not necessarily vastly less than traditional methods of measuring for a typical building survey, the quantity of data and detail captured affords huge gains in building insight in the office. Much more detail can be shown in a survey drawing and coherent 3D models can be constructed with relative confidence. In terms of the deliverables of the average laser-scanned versus sketched-and-measured surveys, one may not necessarily be able to tell the difference at first glance – depending on the detail the surveyor chooses to show. However, overlaying a reasonable-quality, traditionally-crafted survey with its corresponding modern scan data will show considerable discrepancies.

It has become apparent while working with scan data over the years how inaccurate traditional surveys actually were/are, with respect, particularly, to the plumbness and trueness of built walls. Delving deeper into the comparative sets of survey drawings you will notice that arcs and kinks in building lines have become more prolific as the detail of actual structures and surfaces can now be expressed more intimately. Also, a detailed picture of levels – floor and ceiling – can be established on a grid at any given interval with ease. When drafting sections, in particular, bowing, sagging, leaning and sloping are all surprisingly common features, even in the most modern structures, and can all be represented accurately. With traditional methodologies, these features would have been averaged-out into straight lines or missed entirely.

In reporting on the differences between the two methodologies, it must be noted that there are companies still employing traditional methods, where investment in scanners is considered prohibitively expensive, or the learning curve for adoption too steep for comfort. Survey companies today are assumed by clients to be using the most modern equipment and practices and that is emphatically not the case. Where these companies are competing on unequal terms with respect to their operational costs and skillsets, while delivering below-par surveys, this creates a poor environment for the clients and reputable survey companies alike. Fees are forced artificially low by tendering, and the quality of surveys consequently forced way below what should be expected in a notionally professional field. Given the unregulated nature of the survey industry, it is for the clients to satisfy themselves at tendering stage that their survey contractors and the survey drawings produced are what they expect. As highlighted above, the quality of the survey product (and how it was produced) may not be obvious until it is too late. Especially where tolerances are tight. Membership of a trade body is not necessarily a guarantee of survey competence – and conversely, not being a member does not preclude a companies’ diligence. Caveat emptor.

Scan data capture is an undertaking achievable with minimal skill and experience. What makes for an accurate, homogenised point-cloud of a whole building is how well the scans are stitched (or ‘registered’) together. And that is a function of well-placed scanner locations (and reference targets, if used) on site and an understanding and competence in use of the registration software.

After processing and registration, point-cloud data is imported into the CAD system and there are plenty of capable software tools available for cropping and slicing the point clouds into manageable portions. These ‘slices’ are traced-over in CAD to build the survey in CAD space. There is no magic button (yet) to convert the point-cloud into a complete survey, although tools are becoming more sophisticated in interpreting points and constructing building features accordingly. However, the office drawing process remains the most time-consuming aspect of creating accurate measured surveys.

There are quirks of scan data that must be understood by the surveyor to eliminate errors in drawing. For example, any scanned points toward the edge of a surface are subject to an averaging of distances between the near and far surfaces, either side of the edge. That’s just how the scanners interpret their distances and must be accounted for when looking at the scan data. This normally manifests as diagonal lines emanating from the edge of a surface which, if not paid attention to, can resemble a solid structure disappearing off at an angle. Other notable traps might include moving objects obstructing scans, shadows of data where features are hidden from scan view, difficulties in reading to shiny or dark surfaces, reflections from anything shiny (including wet surfaces) and refractions through windows/glass, distorting the apparent position of features on the other side.

Drawing linework per-se from point-cloud data is an operation that can be picked-up in a day or two by any computer-confident person – like any number of foreign-based drafting outfits successfully selling their cheap services to British surveyors in a cost-crunched marketplace. But interpreting point-cloud data into a coherent, accurate representation of an actual building, with the technical quality and accuracy expected by a discerning Architect is not so straightforward. Just as with traditional methods, a lack of survey experience and knowledge held by a draftsman is hard – but not impossible – to spot at first glance. It is useful to remember that CAD operators are not surveyors and surveyors are much more than draftsmen. To be ultra-cautious in checking the quality of provided surveys (but probably worth doing from time-to-time in any case) a spatially-correct Ortho Image of the plan slices could be requested to overlay onto your final survey linework (a reverse of the drawing process). There are several benefits to this as a diligent client:
 – To confirm that the survey (or the relevant part) was actually scanned.
 – To check how conscientiously the draughtsman has adhered and interpretted the scan data.
 – To assess how well the scan data was registered (double-lines shown in scan slices indicates poorly registered scan data – do you or the draughtsman know which line is correct???)

A quick note on 3D modelling, the qualities of which need to be more widely understood. (Some common misunderstandings of the meaning of BIM will be left aside here. Suffice to say that that a 3D model per-se does not constitute BIM – a 3D model is just a 3D model.) However, it is important to understand what can – and cannot – be achieved with modern 3D modelling software in relation to as-built surveys. The 3D software packages available and prevalent in the industry today are fantastically powerful tools. In design terms they provide neat detailing, depth of content, clarity of concepts and impressive visual impact. What they are not so good at is representing actual buildings as they actually are. The curves, kinks, bows, sags and slopes I mentioned earlier are not so straightforward to incorporate. It is, of course, notionally possible to represent them – the software is mostly sufficiently capable. But the time required to construct these awkward features coherently in 3D can be huge – even for the expert operator. And that is why this level of attention is not normally provided – and certainly not at the prices that clients expect to pay. If pretty, rendered approximations are important to you then a 3D model can go a long way to impress. However, if accuracy is important in your survey, then point-cloud-based 2D plans, elevations and sections are unarguably by far the most reliable and accurate option at accepted price points.

So, the measured building survey industry today, like most other professions, is populated with the good, the bad and the downright inaccurate. As a client, it is straightforward to assess what you’re paying for if you ask the right questions:

• Can your survey service deliver what I need? (what, exactly, are your needs?)
• What is your data-capture method?
• How much experience do your data-capture and data-processing operatives have?
• Who is drawing the survey?
• Who is overseeing the deliverables?
• What guarantees do you provide for accuracy (RICS/CICES accuracy bands?)i

Callidus Building Surveys Ltd – the expert Measured Building Surveyor for London, The Home Counties and beyond.