April 5, 2023
CoGri Group’s Managing Director Kevin Dare takes a close look at the limitations of 3D floor scanning and investigates mysterious circles on concrete warehouse floors.
These look very much like crop circles, but instead are on concrete warehouse floors.
Don’t worry, these do not actually exist on the floor, these are visual indicators of the uncertainty (inaccuracy) of 3D scanners when used to check floor surface regularity (flatness and levelness).
The more you look at a contour plan of a 3D scan of a floor, the more circles you will find.
When carrying out a 3D scan of a concrete floor there is an area directly under the scanner that cannot be picked up, a blind spot. Depending on the height the scanner is set (the higher the better when scanning a floor), the bigger the area under the scanner that cannot be seen. This circle of missed data will need to be filled in by data collected from adjacent scan positions.
The make and model of the scanner, the data collection setting on the scanner and the method of registration, will determine the accuracy of the survey data collected. If the above results in very accurate and repeatable data, the area of missing data will be filled in accurately and you will see no resultant circle.
The circle is more pronounced when the scale of the contour lines is set close together. As the contours are stretched, the visual impact of the circle is reduced, until the circle visually disappears. When all the circles of a given area disappear, this is the point where we could assume this to be the level of uncertainty.
The larger the survey area and the greater the number of scanner locations that need to be married in, the greater the uncertainty may be.
Scan 1 – was carried out by Face Consultants Ltd (UK).
We have carried out an exercise on a strip of floor that is 43 metres long and 5 metres wide from a survey of a floor panel of 2400 square metres.
The data collected was with a Leica P40 with setting 3.1mm at 10m distance. Face Consultants positioned the 3D scanner at maximum 10 metres centres and set at a height of 1.8m. Registration was carried out using Leica Cyclone with a combination of 4.5” black and white scan targets and cloud to cloud being used to align the scans. Scan targets were set with a Leica MS60 total station.
The program used for analysing the point cloud data was N4ce.
The floor was cleaned with an industrial scrubber dryer before the survey.
Scan 2 – was carried out by another survey company.
The data was provided to Face Consultants for evaluation in an E57 format. Face Consultants put this data through N4ce in the same manner as above.
The data collected was from a Faro S70 with an unknown data collection setting at 10m scan centres and set at an unknown height. Registration we assume was with Faro ‘Scene’ and cloud to cloud.
We do not believe scan targets were used.
On each floor area, we have shown contour settings starting at 0.1mm and then steadily increasing by an additional 0.1mm. We can estimate the uncertainty value by looking at the contours around the circles until they all visually disappear. Once there is no evidence visually of a circle, this could be regarded as the uncertainty value of that particular survey, using that given equipment, and employing that specific method.
As can be seen from the contour plans of the two floors, we have an assumed uncertainty value of between 1.5mm and 2.0mm for Scan 1 and between 2.3mm and 3.0mm for Scan 2.
This data would then be used to determine compliance with a given surface regularity specification, which will invariably involve the differences in elevation of two or even three individual points and each point having the above uncertainty value. This concept of testing floors for contract compliance is exasperated by the fact that most surface regularity specifications have limits that are relatively small, and the above uncertainty values could be more than 50% of the limit we are trying to measure.
The resultant effect of a survey method that has such a large uncertainty is a report that will have many errors. These could be positive errors that will be seen in the survey results, but may not actually exist, and negative errors that do actually exist, but are not shown and will be missed. This becomes an issue if the floor is required to be remediated, what do you grind? We have also seen that when the same floor is re-surveyed, new errors are found but in different locations.
3D scanning is not yet a suitable method of determining floor surface regularity compliance for most specifications, especially where the uncertainty value is greater than 10% of the property limits we are checking against and the results determine whether a contractor gets paid or not.
"Thank you for the results, which certainly make for interesting..."
Senior ManagerLeading Manufacturer (Leisure Industry)
"Just a note to thank you and your colleagues for..."
Project Manager McElroy Associates
"Excellent work as always Duane."
John Phoenix, Senior Contracts..."
Senior Contracts ManagerABS Brymar Floors Ltd