A novel approach to high-pressure gas releases simulations

Highlights

• A novel CFD approach is proposed for accident consequences estimation.

• The approach is suitable for congested environments (e.g. O&G offshore platforms).

• The novel approach models accidental high-pressure gas releases via CFD.

• It allows obtaining accurate and fast results, useful in the plant design phase.

Abstract

Risk-relevant plants like Oil & Gas (O&G) or nuclear ones are subjected to strict safety regulations. Risk Assessment is mandatory for these plants, and the damage quantification is a crucial step which has to be carefully addressed.

Nowadays the state of practice for consequences estimation entails the use of semi-empirical methods which permit a fast evaluation of the large number of accidental scenarios needed for a Quantitative Risk Assessment (QRA).

However, in case of large and congested industrial environments like offshore platforms or equipment inside nuclear primary containment buildings, the aforementioned methods usually lead to an overestimation of the damage areas, for example because they neglect the space congestion which highly affects the accident evolution. A more accurate analysis can be performed using the Computational Fluid Dynamics (CFD), nonetheless its high computational cost represents an important drawback.

In this work a CFD approach, called Source Box Accident Model (SBAM), is presented. It models high-pressure gas releases (>10 bar) in congested environments guaranteeing a good computational cost-accuracy compromise. The aim of SBAM is to permit a fast consequences estimation (evaluation of flammable/toxic areas, etc.) via CFD, in order to have a simulations time compatible with the plants design schedule. The long-term objective is to integrate the CFD contribution in a safety driven design process.

SBAM is based on the splitting of the multi-physics and multiscale phenomena characterizing the accident: the initial supersonic compressible release and the successive low speed dispersion. The first one is simulated in a small domain called Source Box (SB) and the second one in the case study domain.

The two simulations are coupled in a suitable way, through proper parameters which are extensively discussed. This work presents a detailed description of SBAM and two different analyses: a sensitivity study on the coupling parameters and a numerical benchmark which uses a standard CFD simulation as reference. The sensitivity analysis shows that the coupling is a crucial step of the method and the coupling parameters must be treated in the most accurate way. The numerical benchmark shows that SBAM is not introducing significant errors with respect to a standard CFD simulation and, in addition, permits a relevant simplification in the simulation setup and computational cost reduction.

Introduction

The field of consequences assessment and damage quantification are crucial aspects in the industrial risk assessment, especially for those plants which are considered risk-relevant like Oil & Gas (O&G) or nuclear ones. In this plants, many activities are performed and involve hazardous substances at high pressure inside equipment and pipes (Paté-Cornell 1993; Necci et al., 2019) e.g. the most recent design of fusion installations foresees the possibility to use high-pressure gaseous coolants (Zappatore et al., 2020). Moreover, consequences are often worsened by the congested environment that additionally challenges the management of the accidental scenarios (Uggenti et al., 2017).

In the framework of safety assessment of offshore installations, the European Union (EU) Offshore Safety Directive 2013/30/EU affirms that Quantitative Risk Analysis (QRA) tools are the most appropriate for damage estimation of major hazards, in particular fire and explosions.

Nowadays, the major part of the consequences analysis for industrial applications is performed through semi-empirical models such as turbulent free-jet models (Chen and Rodi 1980), gas dispersion models (Davidson 1989) and jet fire models (TNO 2005; Zamejc 2014). These tools, based on massive measurements and few theoretical principles, neglect the geometry and approximate the physical phenomena. On the one hand these assumptions guarantee the semi-empirical models’ simple and rapid implementation, essential for the simulations of a huge number of accidental scenarios, typical aspect of a QRA. On the other hand, the previous hypotheses lower the accuracy of the results, leading to consequences overestimation and, accordingly, to overprotected systems, overloaded bearing structures and waste of materials and money especially if highly congested plants are considered. Typical examples are represented by nuclear power plants, since, for safety reasons, the major part of the equipment is in the containment building and O&G offshore platforms, in which all the equipment must be arranged in a small volume due to space issues.

The alternative to the empirical methods is represented by more accurate models, for example based on the Computational Fluid Dynamics (CFD). They guarantee more realistic results, taking into account the complex geometries and phenomena overlooked by the semi-empirical models. As a drawback, the computational effort and the implementation complexities limit the use of these tools during the preliminary phases of the design and become incompatible with the project schedule (Impalà et al., 2017). Usually, then, only few scenarios are CFD analyzed, and the results are available when the key design choices are already made, reducing the CFD role to a final verification.

To overcome this drawback, this work presents a new approach, called Source Box Accident Model (SBAM) (Carpignano et al., 2017), targeted to be a compromise between the semi-empirical and pure CFD models and aiming at decreasing the calculation costs maintaining a good results accuracy.

The approach is focused on toxic and/or flammable accidental gas releases from pipelines or tanks in complex geometries. Additionally, this tool can be used also for asphyxiating gases (e.g. CO₂). Due to the frequent presence of high-pressure components in industrial plants, accidental releases from 10 bar or more are considered. This phenomenon is complex being characterized by a multi-physics and a multiscale nature: the jet is supersonic and compressible near the release point and becomes a subsonic and incompressible dispersion in the remaining portion of the domain, the largest one.

From a literature survey, resulted that several papers treated the CFD simulation of gaseous releases and dispersions. Some papers are focused only on supersonic free-jets (Birkby and Page 2001; Fairweather and Ranson 2006), and do not account for neither the jet-objects interaction nor the gas dispersion. On the contrary, in (Lin et al., 2021; Li et al., 2019; Wang et al., 2020; Dong et al., 2017) only the gas dispersion due to a low-inertia release is modeled. More complex high-pressure releases in big environments are treated in (Venetsanos et al., 2008; Choi et al., 2013; Liu et al. 2014, 2015; Deng et al., 2018); however the initial supersonic release is never described by CFD tools, but it is substituted by semi-empirical approaches or other simplified models based on conservation laws. For example, two of the most used CFD software for QRA, FLACS (Gexcon 2019) and KFX (DNV GL 2019), use the Birch model (Birch et al., 1984) to evaluate the initial jet-expansion, and a simplified formulation of the flow field governing equations (Porosity Distributed Resistance CFD approach). In (Wilkening and Baraldi 2007) both release and dispersion are simulated in a single CFD simulation, nonetheless the resulting computational cost is prohibitive and a High Performance Computing is employed. SBAM is developed to handle both the supersonic release and dispersion via CFD in a detailed and computationally sustainable way.

In section 2 the problem statement is introduced by describing the main features of the physics involved in the phenomena treated by SBAM, with particular attention to highly under-expanded jets; two different flow regimes are identified: the supersonic release and the low velocity dispersion.

In section 3, the rationale behind SBAM is explained. The concept of Source Box (SB) is introduced, i.e. a small domain in which the release phase is simulated. Then, it is explained how the dispersion is simulated taking as input the results of the SB. Hence the coupling of these two simulations is described.

In section 4, the two analyses carried out on SBAM are described: a sensitivity analysis on the coupling parameters and a numerical benchmark.

In section 5, a typical case study, which can be treated by SBAM is introduced: an accidental high-pressure gas release in an offshore platform. The numerical methods employed in the SBAM simulations are discussed.

In section 6, the results are discussed with reference to the typical SBAM output: damage areas, damage metrics as flammable volume and mass.

In section 7, final considerations and conclusion on the model are summarized.