Securiton AG, based in Zollikofen/Berne, Switzerland, part of the Swiss Securitas Group, is a manufacturer and installer of high-end security and fire detection systems. Securiton has a global presence in fire detection systems for special applications such as Line Type Heat Detectors and Aspirating Smoke Detectors (ASD).
Background
An ASD is based on a sampling pipe network with a number of sampling points, there is an evaluation unit with a fan and a high sensitivity smoke sensor. Air samples are taken from the sampling points and checked by the smoke sensor. In case of smoke, the ASD signals a pre-alarm and a fire alarm to a fire detection system. An airflow monitoring system checks the airflow for tube blockage and tube breaks. Most ASDs are designed to European Standard EN 54-20 which has three application Classes A, B and C.
The Challenge
ASDs are used for room protection applications (class C) similar to the well-known point type smoke detectors. Modern ASD systems are able to provide up to 50 or more sampling points with sampling tube lengths of up to 300m. Securiton AG has to ensure that the detection behaviour for every sampling point equates to point type detectors taking into account different topography and distances of the holes.
The physical location of the sampling holes is governed by the area to be protected. The sampling tube topology has to be optimized to provide the smallest sampling tube length and shortest sampling times.
The Solution
To solve this problem, Securiton developed the ASD PipeFlow calculation tool to assist in the development of ASD projects. A graphical interface provides a modelling tool to specify the sampling tube lengths and sampling point positions. The ASD PipeFlow also calculates the necessary sampling point diameters, resulting balanced sampling times from every hole, resulting pressure at every hole and necessary detector sensitivity setting to meet the application class criteria in the EN 54-20 standard and the requirements of NFPA 72.
Securiton turned to NAG to provide numerical routines to solve the complex equations. They needed assurance that the routines would ‘stand the test of time’ and as their hardware changed they needed to be sure the routines would remain robust and accurate.
For each component inside the pipe system, a non-linear relation between pressure and flow can be modelled and experimentally defined. This then defines a system of dependent, non-linear and transient equations. The system has hundreds of equations and variables to be considered.
During their evaluation of the NAG Library, Securiton experience increased acceleration using the mixed algorithm of Newton and Downhill (found in the Optimization Chapter of the NAG Library). The speed and numerical stability were found to be very good. However, the principal reason for embedding NAG routines in their application was the good convergence behaviour demonstrated, compared with other algorithms tested during the evaluation period.
The pipe system is built up graphically using Microsoft® .NET. When the calculation is prepared this is converted in a tree of functional objects which actually defines the equations. The NAG Library routine is used from C++, using a little managed CLI component, with a call-back function to the equation evaluation in the managed world. It is important that memory is not copied around for the process to be efficient.
A spokesman for Securiton said “The performance gains are really very good. So much so that a system optimizer could be built on top of the described solution finder. Each step of the optimizer will try to optimize the aerodynamic behaviour by a discrete variation of the pipe system components.
In retrospect, using the NAG Library gave us more time to concentrate on the actual problem, and we felt completely assured that using NAG’s routines would give us the numerical solution in an efficient and numerically stable way.”
