Abstract
Smoking flares have always represented the visible symbols of pollution escaping from the world's oil and petrochemical industries. Only in recent years, however, has the increasing size of such complexes begun to cause acute environmental problems. During normal operations, these valuable off-gases ought to be conserved, either by gas liquifaction or by re-injection plants but, in an emergency, flares remain necessities as 'plant safety valves'. Essentially, smokeless flaring is an aerodynamic mixing problem. The large volume air requirement for clean combustion has to be induced into the flame by some kind of high pressure gas or steam entrainment system. A novel method of effecting this mixing, rapidly, is to utilise the 'Coanda Effect'. This is the name given to the tendency of a fluid jet to adhere to an adjacent surface, thereby causing enhanced entrainment, actually up to 25 times the original air inflow into the jet. The object of this research was to study and thus improve the performance of one design of Coanda, external surface, steam flare, developed and marketed by BP under the trade name Stedair. The investigation of this flare system was conducted in two parts; physical modelling of the entrainment at the flare tip, followed on by confirmatory measurements, on a 76 mm pilot scale, propylene doped, methane-steam flare, A slice (two-dimensional) water model of a Stedair flare tip was constructed, to examine the effectiveness of Coanda surface geometry, as an entrainer and mixer. The neutralisation reaction between solutions of HCl and NaOH was used to imitate the combustion reactions between oxygen and hydrocarbons, the neutralisation being made visible by the use of phenolphalcin as an indicator. Adjustment of the concentrations cuid flows simulates fuel stoichiemetry and excess air requirements, producing a 'flare envelope' corresponding to the 'mixed is burnt criterion. The second phase of the investigation consisted of a series of combustion tests on the largest (76 mm) Stedair steam flare that could b6 safely operated within a high ceiling (7.6 m) laboratory. Temperature and CO profiles within the flame were measured on 3 flare heads, with 4 slot widths/head and at 3 steam flows/slot width, the gas flow being kept constant. Smoke Points, Blow-off and Coanda Breakaway Points were also measured at varying gas flows. The Coanda principle is inherently such an excellent entrainment device that combustion on these flare tips is nearly always good, provided that the jet slot width remains uniform. Whilst noise and radiation levels can only be measured on real flares, combustion quality and entrainment rates are best measured on slice water models of such flares. The experimental flare built by the author demonstrated this point very well. The pulsations and oscillations of this turbulent jet, even in the wind-free laboratory environment made obtaining consistent measurements very difficult. A Coanda flare behaves like an enclosed rather than a free jet, thus the similarity criteria developed by Craya and Curtet can be applied and when it is, Jenkins and Coworkers have shown that, water model results do correspond to those measured on real flare trials.