Pressure recovery systems for supersonic chemical lasers:

parameters selection and implementation

A.S.Boreisho, V.M.Malkov, A.V.Morozov, I.A.Kiselev, A.E.Orlov, A.V.Savin, I.V.Shatalov, S.L.Druzhinin

 

Supersonic chemical lasers represent nowadays the most powerful source of continuous radiation. (HF/DF-CCL - continuous chemical laser; HF, DF - hydrogen-fluorine, deuterium-fluorine are the active mediums; COIL - chemical oxygen-iodine laser). Achieved development level of these devices allows setting the problem of independent mobile complexes (IMC) design based on these devices.

Emission of waste active medium into the atmosphere is one of the problems accompanying the IMC design. The resonating laser cavity should contain supersonic flow (н≈2) with static pressure which is significantly lower than atmospheric pressure. Stagnation of such a flow is accompanied by substantial mechanical energy losses, therefore special pressure recovery systems (PRS) are required to provide emission into the atmosphere. Ejector type PRS is an efficient system which, essentially defines the sizes and weight of the whole IMC. In this case pressure recovery up to the atmospheric level can be provided by a single-stage ejector. Alternatively, in case of COIL a two-staged ejector is sufficient. Therefore the problem of efficient PRS development (with minimal weight and size) is one of the key stages of IMC design.

PRS consists of supersonic diffuser (SD), heat exchanger (HE) with vacuum chamber, and ejector (EJ). Besides, some cases require additional elements: sound suppressing and neutralizing systems. As a matter of principle, PRS structure resembles the exhaust systems of supersonic wind tunnels: exhaust diffuser and single- or multi-stage ejector. In aerodynamics SD and EJ performance is estimated using one-dimensional approaches based on integral conservation equations. Experimental data are used to evaluate the lengths of SD and EJ, so the approaches are semi-empirical. However, in case of supersonic chemical lasers (SCL) these approaches almost fail to work.

Firstly, shape of PRS duct significantly differs from the shape of wind tunnel duct (and shape of a duct is an essential feature in gas dynamics): laser duct has a rectangular cross section with large aspect ratio h/b<<1, and aerodynamic duct is generally cylindrical (sometimes rectangular but with an aspect ratio of h/b≈1). And the flow in such extended rectangular duct noticeably differs from the flow in round pipe as 3D effects in the corners significantly influence the flow.

Secondly, laser gas is hot and light (Ю=8-10 g/mole and ф=1100÷1400л) at low pressure unlike the cold air in wind tunnels (Ю=29 g/mole and ф=300л). It means that characteristic Reynolds numbers for SCL flows are as a rule 2-3 orders less than for the pipes (therefore boundary layers are thicker and viscous effects are stronger).

Thirdly, unlike the wind tunnels in SCL gas flow is supersonic with internal heat generation provided by chemical reactions in case of HF/DF-CCL and by thermalization of singlet oxygen molecules in case of COIL.

All these circumstances (duct shape, low Re numbers and heat generation) essentially influence the stagnation process in exhaust SD.

The process in PRS ejector is also atypical for wind tunnels: hot and light gas is ejected and this is the most complicated mode of ejector operation. Besides, different schemes can be employed to minimize the sizes, e.g. multitubular system. The distinguishing features of heat exchangers consist in rather specific parameter range of PRS SCL, and experimental data used for empirical approaches can noticeably differ in works of different authors.

Three-dimensional computations of real viscous flows are required to design new gas dynamics apparatus operating in new parameter ranges (PRS).

Modern computer technologies and numerical methods allow 3D simulation of flows on basis of complete system of Navier-Stokes equations. Commercial programs are available (FLUENT, CFX, etc.) which allow conducting parametric research of viscous turbulent flows and obtaining complete picture of all flow parameters in any point of calculation domain in any point of time. Possessing large set of governing parameters it is impossible to choose optimal geometry, duct operation and start conditions etc. without such calculations. Results obtained with traditional semi-empirical approaches can be applied for choosing the initial geometry of calculation domain and its main parameters.

However, existence of such software does not mean the immediate and correct calculation of the flow. Only simple turbulent flows at low Reynolds numbers can be calculated by direct numerical simulation (DNS), and it requires a huge amount of time. It is necessary to apply turbulence models the variety of which complicates the justified choice. Thus, the successful calculation requires the verification of the chosen calculation approach.

Verification is a complex concept. Not only a suitable turbulence model is required, but also physical phenomena included in computational model should be defined (e.g. gas dynamics equations are supplemented with chemical kinetics equations when the flow in chemical laser nozzle is being calculated). Such complex objects (like SCL with PRS) require some calculation strategy. The problem should be divided into a number of subproblems (domains). In one part of a channel (e.g. in laser chamber behind the laser nozzle unit) chemical reactions dominate in gas flow. In the other part (e.g. in diffuser) gas dynamics of supersonic viscous flow (interaction of turbulent boundary layer with shock waves) predominates the rest of the processes. Although some kinetic processes may also be taken into account (here some relaxation processes occur). Thus, conditions of adjustment and matching the solutions should be obtained and boundary and initial conditions should be correctly set in different domains. The choice of computational domains is often determined by design features of device itself, and the choice of computational grids follows the features of particular flows at the given channel section (this is not confined to automatic grid refinement in boundary layer domain). Grid convergence of solution should be researched.

Thus, verification is a complicated research work which includes not only a huge amount of calculation but also experimental work on model apparatus and subsequent corrections. Experimental testing and confirmation of numerical results can be followed by development of verified calculation approach which allows obtaining the best solutions for the device being created. The described in paper methodology, i.e. numerical parametric research based on verified computational methods, allows implementation of complex gas dynamic devices like PRS for SCL.