Microsoft Word - 35fakandu.docx CHEMICAL ENGINEERING TRANSACTIONS VOL. 53, 2016 A publication of The Italian Association of Chemical Engineering Online at www.aidic.it/cet Guest Editors: Valerio Cozzani, Eddy De Rademaeker, Davide Manca Copyright © 2016, AIDIC Servizi S.r.l., ISBN 978-88-95608-44-0; ISSN 2283-9216 Large Scale Carbon Dioxide Release: Short-Cut Analytical Modelling and Application Emilio Palazzia, Erika Lunghia, Andrea P. Reverberib, Bruno Fabianoa* a DICCA Civil, Chemical and Environmental Engineering Department - Polytechnic School, University of Genoa, via Opera Pia 15, 16145 Genoa, Italy b DCCI –Chemistry and Industrial Chemistry Dept.– Genoa University, via Dodecaneso 31, 16146 Genoa, Italy. brown@unige.it The ongoing development of CCS applications and installations at large scale involves the need of improving the knowledge of connected hazards resulting from accidental loss of containments, or intentional events. In fact, a massive release of CO2 can have catastrophic consequences for humans: the processes determining the hazards posed by accidental releases of CO2 from pressurized systems are complex, due to the thermodynamics of the outflow, with changes of phase, followed by the dispersion of the cold heavy gas. In this paper, we explore a peculiar scenario connected to a massive release of carbon dioxide and following accumulation driven by negative buoyancy effect under semi confined conditions, either due to low wind and natural complex orography, or to the presence of geometrical complications. The paper sets out a preliminary analytical model, developed under simplifying but conservative hypotheses, which can be conveniently adopted at least at the early stage of the evaluation process or for establishing emergency procedures defining critical distances and possible man exposure to the hazardous dose. 1. Introduction Effective modelling of CO2 release situation is essential for pipeline and storage design and safe operation within sensitive or inhabited areas, as well as in obtaining stakeholder acceptance. In the aftermath of severe accidents Regulatory Bodies, research companies, healthy organizations and more generally society are forced to re-examine the way things were done, determine immediate and root causes and make appropriate changes possibly applying novel methodologies and solutions (Vairo et al., 2016). Case histories provide an empirical contribution to our understanding on the hazard distances for CCS projects, also in view of QRA based decision. A well-known accident involving a massive and sudden release triggered by a natural event took place in Cameroon where carbon dioxide accumulated over the years in the lower strata of Nyos lake, was released with low momentum for a total estimated mass of 1.5 million tonnes (Kling et al., 1986). The natural orography of the valley with a flat depression allowed accumulation and transport, causing nearly 1700 fatalities and a large number of killed livestocks. A previous example is provided by the extreme outburst of carbon dioxide occurred in 1953 in Menzengraben evaporate (potash) mine (former East Germany). Under still wind and stable weather conditions an estimated mass of 1100-3900 tonnes gave rise to a high momentum vertical release, with little impingement, and the subsequent CO2 accumulation caused several fatalities and injured people by asphyxiation (Hedlund, 2012). Different theoretical studies focused on simulating the release and dispersion of dense phase CO2 from high-pressure media (Witlox et al., 2009) using CFD and hazard analysis software tools (e.g. Dixon et al., 2012). Moreover recent experimental studies still under development are addressed at experimentally verify all relevant physical aspects in order to develop and validate mathematical models for discharge and dispersion from dense-phase CO2 pipelines (Jamois et al., 2014) and connected loss of containment frequency (Milazzo et al., 2015), or storage sites. For instance, the Shell Barendrecht carbon dioxide sequestration project foresees that the compressed gas would be stored in an empty natural gas cavity beneath the town of Barendrecht (NL). In this case, although in the Netherlands an individual risk criterion of 10 -6 at contours around a static risk source is well established, the QRA-based decision was overruled. However, when conservative results are enough as a first screening tool, analytical DOI: 10.3303/CET1653060 Please cite this article as: Palazzi E., Lunghi E., Reverberi A., Fabiano B., 2016, Large scale carbon dioxide release: short-cut modelling and application, Chemical Engineering Transactions, 53, 355-360 DOI: 10.3303/CET1653060 355 models can be conveniently applied in hazard assessment (Fabiano et al., 2015), to evaluate hazardous maximum build-up (Palazzi et al., 2013), or to evaluate hazardous events posing a higher risk than the safety level and to determine safety measures (Abrahamsen et al., 2013). The objective of the current work is to develop a short-cut model for predicting on the basis of few parameters, the scale and extent of the hazardous area for the most sensitive receivers, without accounting on CFD models needing proper accurate set-up and long computational time (e.g. Basso et al., 2015). 2. Modelling framework Figures 1 a-b-c describe the simplified time evolution of a quasi-instantaneous carbon dioxide release, under still wind conditions. A dense cloud of volume V0, having similar vertical dimension, h0, and horizontal dimension, 2 r0 , near the source is formed, it is subject to slumping by gravity and it is diluted with air as it expands radially. After a rapid initial dilution, completing the sublimation phase of carbon dioxide, the initial momentum of the release is exhausted, because of its higher than air density and its friction with the ground. Schematically, the dense cloud slumps under the influence of gravity while increasing its radius, r, and reducing its height, h. The evaluation of the total mass of the cloud, after the jet phase and at the beginning of the slumping phase, m0, is performed by assuming the proportionality between the mass of entrained air and the initial release momentum flux. The subsequent spreading neglects the influence of atmospheric turbulence and wind, as a first simplified approach and relies on the non-dimensional density definition introduced by van Ulden (1974). Under the assumptions of flat terrain, no obstructions, no local concentration fluctuation, no chemical reaction, the unsteady – state behaviour of a nearly instantaneous carbon dioxide release, near the ground, can be described by following Eqs (1)-(4): dVdt = kπr v v = drdt v = 1r ρ − ρρ gVπ V = πr h (1) (2) (3) (4) Eqs (1)-(4), allow determining the characteristics of the cloud depending on the horizontal dimension, r, and/or time ,t, once given the value of the mass release at the end of the jet phase, mr, By combining Eq (1), which defines the air entrainment speed within the cloud and Eq (2), providing the cloud slumping speed, with subsequent integration one can write: V = V + 13kπ(r − r ) ≅ V + 13kπr (5) Starting from similar studies (Webber, 2011), we considered the following 4 reference operative situations for storage/transport (respectively I-IV), namely: Ti= 273 K and pi = 100 bar; Ti= 273 K and pi = 200 bar; Ti= 323 K and pi = 100 bar; Ti= 323 K and pi = 200 bar. We explored the six environmental conditions schematized in in Table 1, thus obtaining 24 reference release scenarios. Figure 1 a-b-c : Physical model of a nearly instantaneous carbon dioxide release and time evolution. 356 Table 1: Explored environmental conditions For the purposes of dose calculation, it is fundamental to determine the post release variations of carbon dioxide concentration y(r), during the cloud slumping. Starting from the jet modelling developed in Palazzi et al., 2016, it can be assumed: V = mρ = mρ w (6) V − V = mρ 1w − 1w (7) By combining Eqs (6) and (7), the correlation between the volume, V, and the released mass, mr, can be obtained: V = α m (8) where the parameter αv [m 3 kg-1] is provided by: α = 1ρ w + 1ρ 1w − 1w (9) Combining Eq(5) and Eq(7): r = α m (10) where the parameter αr [m kg -1] is provided by: α = 3kπρ 1w − 1w (11) From Eq(4), it follows: h = α m (12) where the parameter αh [m 3 kg-1/3] is provided by:: α = αvπα (13) The carbon dioxide concentration during the cloud slumping, y(r), can be obtained with some straightforward calculations as follows: y(r) = 1 + MM 1w − 1 + 13kπρm r (14) The simplified description of the cloud evolution as time goes on is performed according to the framework outlined in the following. Taking into account Eq(6), Eq.(3) can be conveniently written as: v = 1r ρ − ρρ gπ mρ w = βr m (15) where the parameter βi [kg -1/2 m2 s-1] is provided by: β = (ρ − ρ )ρ gπw (16) By integrating Eq.(15), one can write: r = r + 2β m t ≅ 2β m t (17) 1 2 3 4 5 6 T [K] 273 273 298 298 323 323 yw 0 0.006 0 0.0313 0 0.1216 357 At last, from Eq(17) it is possible obtaining the function r(t) and, if needed, v(t), h(t) and y(t). As amply known, in order to assess the CO2 toxicity it is necessary to calculate the exposure conditions in terms of concentration and exposure duration. Given a certain value of the critical concentration yc, the corresponding average concentration within cloud, yL= , is attained at the time tL, when, owing to slumping, the cloud extension reaches the distance rL. In other words, we cautiously evaluate the dose, D∞L, considering that in each point of the circumference having radius equal to rL, the concentration is zero when t