Skip to main content

Factories of the Future Resources, Technology, Infrastructure and Services for Simulation and Modelling 2

Periodic Reporting for period 2 - Fortissimo 2 (Factories of the Future Resources, Technology, Infrastructure and Services for Simulation and Modelling 2)

Reporting period: 2017-08-01 to 2018-12-31

Fire and Gas detectors are commonly installed in manufacturing and production facilities utilising hazardous processes to automatically alarm and trigger safety measures in response to hazardous events. Without effective detection, installations can be susceptible to three main hazards: (1) accumulation of toxic gases to levels that exceed given exposure threshold limits, (2) accumulation of flammable gases to levels that can cause fires or explosions and (3) various types of fire events.
The design of gas detection systems is currently based on fairly simple, often company-specific standards or rules, and carried out with ‘mapping’ software. It maps each detector within a defined volume of specified size, in which the detector is assumed to detect all possible leaks. This method, while simple and robust, can lead to excessive detector numbers if the hazards are not well understood, and the parameters to be used in design (such as the cloud size required to cause damaging overpressures) are selected poorly. Also, this approach is not based in any way on the typical wind fields or actual flow patterns in the facility.
Based on its computational fluid dynamics software FLACS, Gexcon suggested the development of a new application in gas detection system optimization, based on running many different gas dispersion scenarios. While experienced with traditional ‘mapping software’, Micropack’s intention to join the experiment has been to be involved in the development of the next generation gas detection system optimisation methodology, based on CFD simulation.
The objective of this experiment has been to improve the installation design process by using advanced CFD simulation and modelling techniques in a cloud-based framework, which exceeds standard modern ‘mapping programs’ for gas detection purposes. The solution provides a firm and repeatable base upon which complex reasoning for designing detector systems should be based. It provides improved safety and reduced detector installation and operating cost in a vast number of industrial/manufacturing facilities that incur risks of gas releases, explosions, fire, etc.
Since the computational requirements of this system are considerable (simulation of many different scenarios), the solution is provided as an HPC-cloud based simulation and modelling service.
The main technical work items of the experiment have been:
• Implementation of the technical infrastructure for GUI-based submitting and monitoring of simulation jobs on the HPC system.
• Integration of the GUI functionality in FLACS-Risk, as the first HPC client.
• Implementation of new tools for CFD-based detection optimization. This is a new application area that requires high numbers of simulations (100s—1000s) and has not been possible to tackle without the new functionality and HPC resources. As part of this development, validation and methodology development has been started.

The important result on the business side is the investigation of future business models for FLACS. Gexcon currently sells its software based on different license types (based on application areas and duration). Changing to a pay-per-use model could increase the number of customers, but also poses a significant risk to the company’s income base.
Running many CFD scenarios for detection optimisation or general risk studies in the HPC-cloud shortens the time scale from weeks to days, which is in accordance with analysis and planning cycles for the type of installations considered. FLACS users further avoid the investments into computing hardware, and in the future possibly also the overhead of the high license cost; the business model for the FLACS-Cloud still needs to be finalised, but in any case, consulting companies will have more options than traditionally and will be more competitive with the new scheme.

In the existing HPC offering, the customer must hold a license for the FLACS software (product of the project partner Gexcon) to obtain access to the HPC system, which serves as extra computing capacity to run many simulation jobs at peak times. For the first phase of the FLACS-Cloud service it has been determined that the same condition will apply. This means that the extra support is limited to technical issues regarding the service, as opposed to allowing first-time users straight onto the cloud service, which will in addition lead to lots of questions about basic FLACS usage.
Gexcon has the following plans building on the experiment results:
• FLACS-Cloud will be released to several pilot customers with the next version of a new product, FLACS-Risk v1.2 in Q1 2018.
• Detection system optimisation as part of FLACS-Risk (tentative: Q2/2018).
As a natural part of the integration in Gexcon’s product portfolio, the new services will be covered also by the course and certification offering.

The new detection optimisation software can provide improved safety and reduced costs for both detector installation and operation (i.e. CAPEX and OPEX) in a vast number of industrial/manufacturing facilities that incur risks of gas releases, explosions, fire, etc. The same holds for other applications of FLACS, which are also possible with the general HPC service.
The improved safety of the system can only be quantified based on statistical data regarding the occurrence of hazardous events. The cost saving of one or several detectors in a detection system quickly adds up to figures on the order of €100,000. Typical gas detection systems comprise several tens of detectors, meaning that overall saving can run to millions of Euros over the lifetime of an installation.
Preparing gas detection optimisation for a small offshore module in FLACS-Risk (prototype).
Gas dispersion simulation results are input for detection optimisation and general risk studies.
Simulations can be run in the HPC cloud from the FLACS user interfaces.