Multi-Point vs Point Cloud

 

 

Detect3D allows you to choose between two methods for calculating coverage for flame detectors: the point cloud method and the multi-point method.

 

The differences in coverage when comparing the point and multi-point methods can be significant as their approaches are fundamentally different. Choosing which method is best for you will depend on the project.

 

Ideally, for fire mapping we want to calculate the response of detectors to flames. Unfortunately, this is much harder than it sounds – to do this accurately would require us to know the full power spectrum from the flame, its modulation, information to account for attenuation and scattering, and finally we would need to know the internal algorithms the detector uses to calculate if a fire is actually there or not (and not a false alarm source), which is proprietary. In other words, what seems simple, is in fact beyond our current engineering knowledge.

 

Furthermore, flame detectors are not tested for obstructed flames. The testing for FM 3260, which certifies ranges of devices, is done in completely unobstructed (ideal) conditions. We did not have any data about how flame detectors respond to obstruction until in 2019 when Insight Numerics and JGC did a joint presentation discussing the first experimental testing of this type at the FABIG TM99 meeting.

 

Considering all of the above, leaves us with quite a difficult situation because there is very little to go on when it comes to calculating coverage for obstructed fires. The best we can do is something sensible, but within that definition there are quite a few possibilities.

 

T he most obvious method is to calculate the region of a zone that is visible to a detector and divide that by the total volume of the zone. This is the “Point Cloud Method" or "Point Method”. The points are simply a discretization of space to help with this volumetric calculation. It is important to note that fires are not considered as points using this method. Points are just dots in space that help calculate the required volumes. We can change the discretization to be more refined (containing more points, e.g. resolution of 0.2 meter spacing) or more coarse (containing less points, e.g. resolution of 0.5 meter spacing). 

 

The downside of the point method is that very small obstructions can cause significant blockages to the detector. It would be impractical if you are performing analysis for a large fire (e.g. 1 meter base) that 1 inch pipes should be considered obstructions.

 

This is where the Multi-Point method comes in. A series of points is defined by a width and height. These points are then group together to represent a region (Multi-Point Region or MP Region) requiring detection. At each location in the zone, Detect3D evaluates how many of the points in the MP Region are visible to the detector. One application for using this is to use the Heskestad Flame Height Correlation to calculate a height of a region which requires detection. ??***

 

One application of this multi-point method would be to have the MP Region represent a flame, although other applications exist.

 

Once the dimensions of the MP Region is defined Detect3D then calculates the number of points within the grouping which can be "seen" by one or multiple detectors. 

 

Using the experiemtal data from the FABIG TM99 presentation as a basis, Detect3D determins the percent of the region needed to be "seen" by the detector to go into alarm based on how much of the region is blocked. The result is that closer the MP Region is to the detector, the less of it needs to be visible because, like a fire, the intensity received by the detector will be greater. 

 

The below diagrams describe the application of the multi-point method. This is chosen by change the Method of the zone to Geographic: Multi-Point

 

Changing the coverage calculation method for an already defined zone to the Geographic:Multi-Point

 

 

Using the Multi-Point method, the size (width and height) of fire to be detected is entered. At the detectors maximum range 100% of the MP Region must be viewed for the detector to go into alarm

 

 

 

If the flame detector is closer to the fire then by the inverse square law, not as much of the box is required to be seen for the detector to go into alarm

 

In addition to the above, users may enter a static value of acceptance criteria for the amount of the MP Region that needs to be seen for the detector to go into alarm. This is regardless of the distance the detector is away from the MP Region. To use this option, select the Geographic: Multi-Point (Fixed) as the calculation Method and enter a coverage percent for the amount of the box that needs to be seen.

 

Which method to choose depends on your project. In general, it is good practice to start out with the Point Method. However, if the coverage is lower than expected, and the fire you are considering is much larger than many of the obstructions on the site, then the Multi-Point Method may be helpful. 

In general, the Multi-Point Method leads to better coverage, however there are cases where the reverse is true. The following images are from an example project and the resulting 0ooN isovolumes and coverages for each calculation method.

Original layout with cuboid zone defined and 5 flame detectors, contour at 1m height is also shown

 

0ooN isovolume for point cloud method with 0.25m spacing of points, this is the default method used when adding zones in Detect3D

 

 

Please contact us at info@insightnumerics.com for more information.