Oxyfuel Optimization using CFD Modeling Thomas Niehoff 1 , Sreenivas Viyyuri 1 1 The Linde Group, Linde Gas, Carl-von-Linde-Strasse 25, 85716 Unterschleissheim, Germany Keywords: Oxyfuel, CFD, furnace optimization, modeling, burner, emissions, heat transfer, Abstract Before converting production furnaces to different combustion technologies it is essential to understand all related changes and side effects. An experienced team will be able to successfully conclude a conversion like this. However, CFD modeling will enable to make informed decisions in terms of effort and results of furnace retrofitting with new combustion equipment. This paper will give insight of how oxyfuel together with CFD can impact energy balance and productivity of production furnaces. Introduction Aluminum recycling and re-melting is a very competitive industry area. Global markets and globalization of aluminum melting technologies and aluminum trade brushes up the dust in every corner of the business. Aluminum producers with large melting furnaces are constantly under pressure to bring production cost down and hence to use the latest available technology. Any change is associated with risks. Furnaces with larger production capacities face higher risk as compared to small ones. Being able to estimate process changes before capital money is spent allows to even optimize a technology towards temperature and heat profiles as well as productivity and energy usage before the system is build. Basic Elements A melting operation consists of various specific elements that need to be transferred into the computer model. However not all can be transferred and not all can be modeled. Hence the model will only be a model and not real. Aspects like charge material quality, quantity, composition and mixtures are very difficult to describe in a model. Charging material storage and altering process conditions as well as the melting itself with all changing physical properties are hard to specify and model. The combustion space of a furnace can be modeled with justifiable effort. The system of the combustion space then needs to be defined and described well enough to come to reasonable results that can reflect the reality. The model will preferably describe steady state conditions, i.e. altering firing rates and flame shapes cannot be characterized in one single step. The model needs to be connected to reality and will need to be verified with the current operation. FLUE GAS SCRAP TYPES AND GEOMETRIES 1 2 3 … n BATH CONDITIONS TEMPERATURES, etc.. GEOMETRIES OF FURNACE AND BURNER PHYSICAL PROPERTIES BURNER FUEL OXIDIZER FIRING RATE FIRING CONDITIONS Figure 1: Basic elements of aluminum melting process. Model Set Up The melting or heating operation to be modeled typically consists out of a specific furnace geometry (rotary furnace, reverberaotory furnace, tower furnace, shaft furnace and many more). This furnace geometry together with refractory material, flue openings, burner positions and geometries, bath level and other protruding elements define the combustion space. Geometries, thicknesses, and physical properties of the materials then are put together to the model. The mesh size of the model describes how detailed the combustion process will be described in a specific location of the model. The mesh size can vary across the furnace. Typically it needs to be very small (detailed) where rapid changes in either geometry or chemistry is expected. The mesh is refined near the region of the burner to capture the gradients effectively. Homogenizing with slow dynamics areas will have wider mesh sizes, when there is an expectation of reduced activity. A typical aluminum reverberatory furnace model with a capacity of 30 t and a footprint of 450 sqft (50 m2) will have 500.000 knots with an average distance of ½ foot (0.15 m). Where each knot represents that all equilibrium equations (heat, energy and mass) will be solved. The model works its way through the furnace system by moving from knot to knot. It can take days or weeks to simulate one single steady state operating point in a way that the results make sense and are good enough for verification. Recent advancements have enabled to solve the governing equations in parallel using multiple CPU’s which reduces the computing time significantly. Typical combustion systems involve high speed flows. Hence choice of proper turbulence models is very important to predict the solution accurately. In most of the cases, reaction is mixing controlled hence turbulence chemistry interaction needs to be 1185 Light Metals 2011 Edited by: Stephen J. Lindsay TMS (The Minerals, Metals & Materials Society), 2011