Prediction of ignition limits with respect to fuel fraction of inert gases. : Evaluation of cost effective CFD-method using cold flow simulations

• Main Author: Sjölander, Johan
• Format: Others
• Language: English
• Published: Umeå universitet, Institutionen för tillämpad fysik och elektronik 2015
• Subjects: Combustion; gas turbines; CFD
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-109094

Improving fuel flexibility for gas turbines is one advantageous property on the market. It may lead to increased feasibility by potential customers and thereby give increased competiveness for production and retail companies of gas turbines such as Siemens Industrial Turbomachinery in Finspång. For this reason among others SIT assigned Anton Berg to perform several ignition tests at SIT’s atmospheric combustion rig (ACR) as his master thesis project. In the ACR he tested the limits for how high amounts of inert gases (N2 and CO2) that the rig, prepared with the 3rd generation DLE-burner operative in both the SGT-700 and SGT-800 engine, could ignite on (Berg, 2012). Research made by Abdel-Gay and Bradley already in 1985 summarized methane and propane combustion articles showing that a Karlovitz number (Chemical time scale/Turbulent time scale) of 1.5 could be used as a quenching limit for turbulent combustion (Abdel-Gayed & Bradley, 1985). Furthermore in 2010 Shy et al. showed that the Karlovitz number showed good correlation to ignition transition from a flamelet to distributed regime (Shy, et al., 2010). They also showed that this ignition transition affected the ignition probability significantly. Based on the results of these studies among others a CFD concept predicting ignition probability from cold flow simulations were created and tested in several applications at Cambridge University (Soworka, et al., 2014) (Neophytou, et al., 2012). With Berg’s ignition tests as reference results and a draft for a cost effective ignition prediction model this thesis where started. With the objectives of evaluating the ignition prediction against Berg’s results and at the same time analyze if there would be any better suited igniter spot 15 cold flow simulations on the ACR burner and combustor geometry were conducted. Boundary conditions according to selected tests were chosen with fuels composition ranging from pure methane/propane to fractions of 40/60 mole% CO2 and 50/75 mole% N2. By evaluating the average Karlovitz number in spherical ignition volumes around the igniter position successful ignition could be predicted if the Karlovitz number were below 1.5. The results showed promising tendencies but no straightforward prediction could be concluded from the evaluated approach. A conclusion regarding that the turbulence model probably didn’t predict mixing good enough was made which implied that no improved igniter position could be recommended. However by development of the approach by using a more accurate turbulence model as LES for example may improve the mixing and confirm the good prediction tendencies found. Possibilities for significantly improved ignition limits were also showed for 3-19% increase in equivalence ratio around the vicinity of the igniter.