Methodology for the estimation of cylinder inner surface temperature in an air-cooled engine

Accepted Manuscript Title: Methodology for the estimation of cylinder inner surface temperature in an aircooled engine Authors: Elisa Carvajal Trujillo, Francisco J. Jiménez-Espadafor, José A. Becerra Villanueva, Miguel Torres García PII: S1359-4311(11)00034-2 DOI: 10.1016/j.applthermaleng.2011.01.025 Reference: ATE 3389 To appear in: Applied Thermal Engineering Received Date: 25 June 2010 Revised Date: 10 January 2011 Accepted Date: 15 January 2011 Please cite this article as: E.C. Trujillo, F.J. Jiménez-Espadafor, J. Becerra Villanueva, M.T. García. Methodology for the estimation of cylinder inner surface temperature in an air-cooled engine, Applied Thermal Engineering (2011), doi: 10.1016/j.applthermaleng.2011.01.025 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. - 1 - Methodology for the estimation of cylinder inner surface temperature in an air-cooled engine Elisa Carvajal Trujillo * , Francisco J. Jiménez-Espadafor, José A. Becerra Villanueva, Miguel Torres García Thermal Power Group, Department of Energy Engineering, University of Seville Escuela Técnica Superior de Ingenieros, Camino de los Descubrimientos, s/n. 41092 Sevilla, Spain. *Corresponding author. Tel.:+34954486111; fax: +34954487243; [email protected]. Abstract A methodology for the estimation of the mean temperature of the cylinder inner surface in an air-cooled internal combustion engine is proposed. Knowledge of this temperature is necessary to determine the heat flux from the combustion chamber to the cylinder wall. Along with the heat transfer coefficient this parameter also allows almost 50% of engine heat losses to be determined. The temperature is relatively easy to determine for water-cooled engines but this is not in the case for air-cooled engines. The methodology described here combines numerical and experimental procedures. Simulations were based on FEM models and experiments were based on the use of thermocouples and infrared thermography. The methodology avoids the use of data or correlations developed for other engines, providing more reliable results than extrapolating models from one engine to another. It also prevents from taking measurements from inside the combustion chamber, reducing invasion and experiments complexity. The proposed methodology has been successfully applied to an air-cooled four-stroke directinjection diesel engine and it allows the cylinder mean inner surface temperature and cylindercooling air heat transfer coefficient to be estimated. Keywords air-cooled engine; heat transfer; cylinder temperature; infrared thermography. - 2 - Nomenclature p j C , specific heat of cooling air at partition i (j = i) or at the outlet fan (j=fan) (J/kgK) i partition (i = 1, 2, 3, 4 correspond to the cylinder and i = 5, 6, 7 to the head) ext cyl h , cylinder-cooling air heat transfer coefficient (W/m2K) m component: cylinder (m=cyl) or head (m=head) exhaust i m , & cooling air exhaust mass flow at partition i (kg/s) & exhaust m m , cooling air exhaust mass flow of component m (kg/s) & fuel m fuel flow (g/s) n number of points of comparison envir p environmental pressure (bar) & cooling m Q , time and spatially averaged heat flow from coolant side wall of the component m to the cooling air (J/s) rejected cyl Q , & time and spatially averaged heat flow through the walls of the cylinder (J/s) & rejected cyl q , heat flux from the combustion chamber and piston to the cylinder inner surface cylinder (W/m2) a R specific gas constant of air (m2/Ks2) s mean square error i S section area of partition i (m2) exhaust i T , cooling air exhaust temperature at partition i (ºC) exhaust m T , cooling air exhaust temperature of component m (ºC) fan T cooling air temperature after fan (ºC) surf measured k T , , measured coolant-side surface temperature at location k (ºC) surf sim k T , , FEM simulated coolant-side surface temperature at location k (ºC) surf inner cyl T , , inner surface cylinder mean temperature (ºC) exhaust i v , cooling air exhaust speed at partition i (m/s) - 3 - [ ] K heat conduction matrix { } N vector of nonlinear heat flows { } P vector of applied heat flux [ ] ℜ radiation exchange matrix abs T absolution temperature { } u vector of grip point temperature 1. Introduction Knowledge of heat transfer in internal combustion engines is important to understand such systems [1, 2] and it contributes to engine development, processes simulation and emissions reduction. In addition, it is important to guarantee the durability of engine components, avoid engine body distortions and improve engine design related to weight and auxiliary energy consumption. Engine heat transfer phenomena have been studied extensively for many decades [3-8]. Numerous mathematical models have been proposed, including correlations based on dimensional analysis, which are widely accepted and, although they propose different heat fluxes, their evolution over the cycle is similar [7-8]. Many of these models include the gas-side wall temperature as a variable to obtain the heat flux through the chamber walls [7]. In addition, computational fluid dynamics (CFD) and/or finite element method (FEM) codes used for heat transfer simulations require the estimation of this temperature to provide boundary conditions. Convergence is attained through an iterative process. [9-10]. Furthermore, thermal analyses require the gas side wall temperature to evaluate temperature distribution and the thermomechanical behaviour of components [11-13]. Other heat transfer analyses use the lumped capacitance method and also need the mean gas side temperature [14-15]. Even with very precise functions, it is not possible to obtain accurate heat flux [7,16] unless accurate wall temperature measurements are available. Some authors approximate the mean temperature of the distribution by one or more –fewlocal measures obtained by thermocouples [1,11,17]. These approaches implicitly assume errors that - 4 - can be acceptable for heat balances but can involve uncertainties in simulation cycles or temperature distribution analyses. It should be noticed that some authors, after having identified one correlation for one small-scale air-cooled engine [18], acknowledge that those correlation parameters are not valid for another small scale air-cooled engine with similar characteristics [19]. In most heat transfer analyses the external surface temperatures are not measured. For example, in the case of water-cooled engines, this temperature is often assumed to be equal to the coolant temperature or is calculated from hypotheses that are specific for water-cooled engines [15] and normally assume constant temperature for all operating points. This approach can not be extrapolated to air-cooled engines because the temperature field in the inner surface changes with the operating conditions [19]. Very few studies concerning heat transfer in aircooled engines have been published and most of them consider two-stroke and/or sparkignition engines [19-21]. In this paper a methodology for an air-cooled four-stroke direct injection diesel engine is proposed. This approach is based on simple energy balances and FEM simulations that use boundary conditions obtained experimentally from thermography, thermocouples and other sensors, while avoiding correlations or experimental data from other researchers. The proposed methodology has been successfully validated, as described in Section 6. 2. Engine description and experimental procedure The experimental investigation was conducted on a four-stroke single-cylinder direct-injection (DI) compression ignition engine. The main characteristics of this engine are given in Table 1. TABLE 1 The experimental measurements covered three engine speeds, namely 1800, 2400 and 2700 rpm, and in each case four different loads were considered – namely 25%, 50%, 75% and 100% of full engine load. Fuel flow, engine rpm and torque were simultaneously measured through all engine load conditions. Environmental pressure and relative humidity were measured using a barometer and a hygrometer, respectively. - 5 - The cooling air is driven by a fan mounted on the flywheel, Fig. 1. A cowling that surrounds the engine fins routes all of the cooling air to an outlet port. An adiabatic plate was installed in order to separate the head and cylinder air cooling flows. The outgoing flow and temperature were measured at 7 points in the discharge area, four of which correspond to the cylinder and the other three to the head, Fig. 2. Air speed, exhaust i v , , and temperature, exhaust i T , , were measured at the centre of each partition “i” with an anemometer/temperature probe. The cooling air exhaust mass flow at partition “i”, exhaust i m , &, was evaluated by Eq. (1). The cooling air exhaust mass flow, & exhaust m m , , and temperature, exhaust m T , , were evaluated by Eqs. (2) and (3). ( ) 5 , , , 10 273 ⋅ =+ &envir exhaust i exhaust i i a exhaust i p m v S R T (1) 4 7 , , , , 1 5 i i exhaust cyl exhaust i exhaust head exhaust i i i m m m m = = = = = = ∑ ∑ & & & & (2) 7 4 , , , , 51 , , , , i i exhaust i exhaust i exhaust i exhaust i ii exhaust cyl exhaust head exhaust cyl exhaust head m T m T T T m m = = == ⋅ ⋅ = = ∑ ∑& & & & (3) The cooling air temperature after the fan, fan T , was averaged from readings from three thermocouples located in the same plane, Fig. 2. Air temperatures, mass flow and heat flow from the coolant side wall of the cylinder to the cooling air for all engine conditions tested were obtained. The main results for the measured air cooling temperature (ºC), air mass flow (g/s) and heat flow from cylinder wall (J/s) are given in Table 2. FIGURE 1 FIGURE 2 TABLE 2 - 6 - Time and spatially averaged heat flow from the coolant side cylinder wall to the cooling air, & cooling cyl Q , , was estimated experimentally from the difference in output and input enthalpy of the cooling air for each operating point, according to the First Law of Thermodynamics, Eq. (4). In this process kinetic energy changes were not considered. The inlet and outlet enthalpy values were obtained from the inlet and outlet temperatures, respectively, where p i C , is the specific heat of air in the grid i of the cylinder, under the conditions considered, and p fan C, is the specific heat of air at the outlet fan, under the conditions considered. The heat losses were evaluated and represented less than 1% of the total. 4 , , , , , , 1 = = = − ∑ && & i cooling cyl p i exhaust i exhaust i p fan fan exhau st cyl i Q C T m C T m (4) Coolant side surface temperature, surf measured k T , , , values were obtained by thermography and thermocouples for all conditions tested. The selection of a greater number of points would be expected to provide better results. If all temperatures are measured with thermocouples the installation would be extremely complex and there would be a risk of failure or the mass flow path could even be affected. As a result, the temperatures were also measured with a thermal camera and this leads to minimal invasion. This approach allows more temperatures to be registered without affecting the cooling air flow. Holes of 20 mm diameter were drilled in the cowling. In order to minimise invasion these holes remained capped when thermograms were not being recorded. Six infrared inspection holes were drilled to register the temperatures of 14 surface locations. Five thermocouples were placed at the bottom fins at the points where the thermograms were unreliable because of the poor angle of vision or the presence of accessories that interfered image acquisition. Therefore, temperatures at 19 locations were measured. The distribution of the holes and the measurement points is represented in Fig. 3. FIGURE 3 Temperature measurement by infrared cameras is a very complex process that is influenced by the environment and the atmosphere. Furthermore, this technique requires parameters that are difficult to measure. In this regard, a set of standards was followed. In order to avoid surface - 7 - temperature measurement errors due to emissivity the contact and non-contact methods were followed [22]. The emissivity was experimentally measured to be equal to 0.85 using a tape with known emissivity, a contact thermocouple and a sample of the metal of the cylinder that was heated in a furnace. The reflector method [23] was used to avoid measurement errors from the reflected infrared energy from heating sources. This method was applied in each engine operating point. During the experiments the most important heat sources that might disturb the experiment, e.g. exhaust duct, were shielded. Standards in [24-25] were used to ensure that the experimental procedure and devices (ThermaCAM PM695) were suitable to obtain the measures. 3. Proposed methodology The methodology proposed to determine the cylinder inner surface mean temperature is based on an inverse method that solves the heat flux components by a steady state heat transfer FEM. The steady state heat balance equation is given by Eq. (5). The equilibrium equation is solved by a Newton iteration. [ ] { } [ ] { } { } { } 4 abs K u u T P N + ℜ + = + (5) One of the main difficulties associated with FEM concerns the correct imposition of boundary conditions [7]. As boundary conditions were imposed the estimated value of heat flow through the internal walls of the cylinder & rejected cyl Q , , fan T , exhaust cyl T , and the convection coefficient on finned surfaces of the cylinder, ext cyl h , . The proposed methodology employed an iterative process to determine ext cyl h , and this converged when temperatures of the coolant side surfaces given by the FEM simulation, surf sim k T , , , matched the measured values, surf measured k T , , , Fig. 4. Thus, ext cyl h , and the inner surface cylinder mean temperature, surf inner cyl T , , , were obtained for each operating point. FIGURE 4 - 8 - 4. Cylinder model The characteristic dimensions of this cylinder are: 95 mm diameter, 180 mm length, 150 mm width and 172 mm height. There are 14 fins of 4 mm width with gaps of 7 mm between each. Several mesh sizes were tested being finally adopted the best compromise between resolution and computational time execution. For example, Fig. 5 show two simple models of one fin with different size mesh. Both FEM simulation results were compared with the known analytical solutions. It was noted that the both absolute and relative errors in both cases very low. Although the errors decreased with the mesh size, this difference is not worth the effort in the meshing and simulation time, Table 3. The maximum size of the element edge is imposed to be 7 mm and there are at least 3 elements of each length. The total number of 3D elements is 54285 (TETRA4) and 19827 nodes and the resolution time is ≈ 500 seconds in a Windows XP based PC, Fig. 6. The material is cast iron with a conductivity of 51.9 W/mK and this remains constant in the temperature range 50-200 ºC. FIGURE 5 TABLE 3 FIGURE 6 5. Application of boundary conditions FEM boundary conditions have to be suitable to ensure goods results. There is a limited set of possible boundaries that can be established using different conditions such as temperatures on surfaces, heat fluxes and convection coefficient. The first attempt involved setting temperatures on the outer surfaces and a suitable heat flux distribution on the inner surfaces. However, the first of the above boundary conditions was impossible to implement due to the lack of information on the temperature field across the entire surface – the temperature of only 10% of the surface was measured. Even in the best case, thermography can only register visible surfaces and the lateral surfaces of the fins were not exposed. An interpolation for each fin, from the extreme temperatures, would be a long and complex process. - 15 - [16] A.J. Torregrosa, P. Olmeda, B. Degraeuwe and M.A. Reyes, A concise wall temperature model for DI Diesel engines, Applied Thermal Engineering 26 Issues 11-12, 2006, 13201327. [17] 17B.R. Tillock and J.K. Martin, Measurement and modelling of thermal flows in an air cooled engine, SAE 961731, 1996. [18] Y. Wu, B. Chen, Heat transfer model for small-scale spark-ignition four-stroke engines, International Journal of Heat and Mass Transfer, Vol 49, Issues 21-22, 2006, 3895-3905. [19] Y. Wu, B. Chen, F. Hsieh and C. Ke, Heat transfer model for small-scale spark-ignition engines, International Journal of Heat and Mass Transfer, Vol 52, Issues 7-8, 2009, 18751886. [20] D. Descieux, M. Feidt, One zone thermodynamic model simulation of an ignition compression engine, Applied Thermal Engineering, Vol 27, Issues 8-9, 2007, 1457-1466. [21] Y. Zhaoda, F. Zheng, Y. Siaoli, Y. Zhixing and S. Hongouan, Calculation and prediction of thermal loading of the air cooled diesel engine, SAE 881254, 1988. [22] ASTM E1933 - 99a(2005)e1 Standard Test Methods for Measuring and Compensating for Emissivity Using Infrared Imaging Radiometers. [23] ASTM E1862 - 97(2002)e1 Standard Test Methods for Measuring and Compensating for Reflected Temperature Using Infrared Imaging Radiometers. [24] ASTM E1213 - 97(2009) Standard Test Method for Minimum Resolvable Temperature Difference for Thermal Imaging Systems. [25] ASTM E1311 - 89(2004) Standard Test Method for Minimum Detectable Temperature Difference for Thermal Imaging Systems. [26] D.C. Montgomery and E.A. Peck, Introduction to Linear Regression Analysis, New York Willey 2º Edition, 1992. [27] Appendix B. Review of probability and statistics, International Geophysics, Vol. 90, 2005, 251-272. [28] D.J. Thornhill, Experimental investigation into the temperature and heat transfer distribution around air – cooled cylinders, SAE 2006-32-0039, 2006. > We propose a methodology for heat transfer calculation. > The analysis is based in cylinder of internal combustion engines. > We combine FEM and thermography technique for surface temperature measurement.> We avoid intrusive technique measurements. Type DI mono cylinder 4 stroke Combustion chamber Bowl in piston Cylinder displacement 708 cm3 Bore 95 mm Stroke 100 mm Compression ratio 19.2:1 Maximum power 11 kW at 3000 rpm Maximum torque 45 Nm at 2100 rpm Fuel Diesel EN 590 Charging Naturally aspirated Lubrication Gear oil pump Injection pump Mechanical Refrigeration Air Table 1. Deutz FL1 906 engine characteristics. load 25% 50% 75% 100% 1800 rpm [54, 101, 2820] [57, 100, 3017] [61, 100, 3251] [64, 100, 3856] 2400 rpm [51, 146, 3170] [60, 151,3562] [57, 148, 4510] [67, 147, 5132] 2700 rpm [47, 175, 3550] [47, 180, 4105] [57, 184, 5489] [60, 184, 6378] Table 2. Results for the operating conditions, [ , exhaust cyl T(ºC), , exhaust cyl m & (g/s), , cooling cyl Q &(J/s)] Node “x” (mm) Analytical T (ºC) Simulation FEM T (ºC) εabs (ºC) εrel (%) 1 0 100.0 100.0 0 0 2 20 75.2 75.4 0.2 0.3 3 40 60.0 60.4 0.3 0.5 4 60 51.9 52.2 0.3 0.6 20 mm mesh size element 5 80 49.2 49.6 0.3 0.7 1 0 100.0 100.0 0 0 2 10 86.2 86.2 0.03 0.04 3 20 75.2 75.3 0.05 0.07 4 30 66.6 66.7 0.06 0.10 5 40 60.1 60.1 0.07 0.12 6 50 55.2 55.3 0.08 0.14 7 60 51.9 52.0 0.08 0.16 8 70 49.9 50.0 0.08 0.17 10 mm mesh size element 9 80 49.2 49.3 0.08 0.17 Table 3. Comparison between known and simulation results for 10 and 20 mm mesh sizes. load 25% 50% 75% 100% 1800 rpm [146, 8.7] [170, 9.6] [150, 12.2] [123, 14.5] 2400 rpm [200, 9.1] [213, 9.5] [233, 12.0] [186, 17.4] 2700 rpm [239, 11.8] [230, 11.7] [250, 11.6] [200, 18.1] Table 4: Results of the iteration process, [ ext cyl h , (W/m2K), s]. Fig. 1. Visualization of air cooling flow and grids (cylinder: 1–4 and head: 5–7). 1 2 3 5 7 cooling air outlet fan cowling cooling air inlet fan head cylinder 4 6 Fig. 2. Experimental set-up. Fig. 3. Cylinder injector side showing the cylinder surfaces that are visible through the holes (white bands) and points chosen for thermographic evaluation (circles). CYLINDER FEM Simulation NO YES rejected cyl Q , & ext cyl h , exhaust cyl T , fan T surf measured k T , , surf sim k surf measured k T T , , , , == surf inner cyl T , , Fig. 4. Iterative process for the determination of cylinder mean inner surface temperature and cylinder-air cooling heat transfer coefficient. Fig. 11. Estimated inner surface cylinder mean temperature.