Tests have been conducted in a wind tunnel with seven types of heat sinks including plate fin, strip fin, and pin fin heat sinks. In the case of strip fin, and pin fin heat sinks, both in-line and staggered arrays have been studied. The pin fin heat sinks had circular and square cross-sections. For each type, tests were run with fin heights (H) of 10, 15, and 20 mm while the heat sink width (B) was kept constant and equal to 52.8 mm. In total, 42 different heat sinks were tested. The width of the wind tunnel duct (CB) was varied in such a way that results were obtained for B/CB=0.84, 0.53, and 0.33. The wind tunnel height (CH) was varied similarly, and data were recorded for H/CH=1, 0.67, and 0.33 while the duct Reynolds number was varied between 2000 through 16500. An empirical bypass correlation has been developed for the different fin designs. The correlation predicts the Nusselt number and the dimensionless pressure drop and takes into account the influence of duct height, duct width, fin height, fin thickness, and fin-to-fin distance. The correlation parameters are individual for each fin design. Further, a physical bypass model for plate fin heat sinks has been developed to describe the bypass effect.

Modeling of the thermal and hydraulic performance of plate fin, strip fin, and pin fin heat sinks-influence of flow bypass

In this study, the heat transfer performance of various commonly used fin geometries is compared. Realistic, manufacturable geometries are considered for minimizing thermal resistance at moderate laminar air velocities and pressure gradients. These consist of plate fins or pin fins, which can be round, elliptical, or square. The plate fins can be continuous (parallel plates) or staggered. The pin fins can be inline or staggered arrays. In order to compare these various geometries, a set of standard conditions was required. The basis of comparison was chosen to be a circular array of 1mm diameter pin fins with a 2mm pitch. The pitch-to-width of the other geometries were chosen to provide equal ratios of fin cross-sectional area to base area. The pitch was fixed to provide the same wetted area per unit volume as that of the nominal case. Analysis was simplified by three major assumptions. Heat transfer and fluid flow were assumed to be two-dimensional, with identical velocity and pressure distribution in the z-direction and similar temperature profiles. Flow and heat transfer were also assumed to be periodically developed, thereby permitting all but the first few rows of fins to be simulated by a model with similar velocity and temperature profiles at inlet and outlet. The ratio of solid to fluid thermal conductivity for aluminum and air is quite high, around 7000, permitting the fins to be modeled as isothermal surfaces rather than conjugate solids. Each of these three assumptions reduced model complexity by about an order of magnitude. The CFD simulations were carried out on a two-dimensional computational domain bounded by planes of symmetry parallel to the flow. The air approach velocity was in the range of 0.5 to 5m/s. comparison of heat transfer performance and pressure drop is presented. In general, the staggered plate fin geometry showed the highest heat transfer for a given combination of pressure gradient and flow rate. This method can be used to compare any repeating fin geometry in a ducted forced convection environment.

A comparison of fin geometries for heatsinks in laminar forced convection: Part I - Round, elliptical, and plate fins in staggered and in-line configurations

The study reported herein extends a previously reported "design for manufacturability" methodology, to forced convection cooled rectangular plate heat sinks. Using a well validated analytical model, the thermofluid performance of the side-inlet-side-exit (SISE) heat sink has been characterized, parametric optimization carried out, and the maximum heat transfer capabilities for a range of operating points has been determined. A least-material optimization has been performed to achieve optimal material use. The analysis indicates the least-material design to provide significant mass savings for a moderate penalty in thermal performance. Empirical criteria for manufacturability obtained from several heat sink manufacturers lead to qualitative guidelines.

Design for manufacturability of SISE parallel plate forced convection heat sinks

In this study, we have attempted to compare the heat transfer performance of various commonly used fin geometries. Realistic, manufacturable geometries are optimized for minimization of thermal resistance at moderate laminar air velocities. The basis of comparison was chosen to be a circular array of 1 mm diameter pin fins with a 2 mm pitch. The pitch-to-width ratio of the other geometries were chosen to provide equal ratios of fin area to base area. CFD simulations were carried out in a two-dimensional computational domain bounded by planes of symmetry parallel to the flow. The air velocity was in the range of 0.5 to 5 m/s. A comparison of heat transfer coefficients and pressure drops is presented.

A comparison of heat sink geometries for laminar forced convection: Numerical simulation of periodically developed flow

Tests have been conducted in a wind tunnel with seven types of heat sinks including plate fin, strip fill, and pin fin heat sinks. In the case of strip fin, and pin fin heat sinks, both inline and staggered arrays have been studied. The pin fin heat sinks had circular and square cross-sections. For each type, tests were run with fin heights (H) of 10, 15, and 20 mm while the heat sink width (B) was kept constant and equal to 52.8 mm. In total, 42 different heat sinks were tested. The width of the wind tunnel duct (CB) was varied in such a way that results were obtained for B/CB=0.84, 0.53, and 0.33. The wind tunnel height (CH) was varied similarly, and data were recorded for H/CH=1, 0.67, and 0.33 while the duct Reynolds number was varied between 2000 through 14000. An empirical bypass correlation has been developed for the different fin designs. The correlation predicts the Nusselt number and the dimensionless pressure drop and takes into account the influence of duct height, duct width, fin height, fin thickness, and fin-to-fin distance. The correlation parameters are individual for each fin design. Further, a physical bypass model for plate fin heat sinks has been developed to describe the bypass effect.

Numerical and experimental studies were carried out to evaluate the air cooling characteristics of plate fins for MCMs (multichip modules) with and without some spanwise space around the plate fins. The flow field in-between the plate fins can be assumed to be similar to that of a parallel plate duct, and hence two dimensional laminar flow analyses using the finite volume method were performed. Calculations were carried out for Res (S/L)=1 to 200 and S/t (plate spacing to plate thickness ratio)=3, 4 and 7. Some empirical equations for the Nusselt number and friction factor based on the numerical results are proposed. Considering the fin efficiency calculated from the conventional conduction fin model, the thermal resistance of the plate fins with no space around the fins could be estimated easily on the basis of the above-mentioned equations. Comparisons with experimental results show that this method is valid for H/S (fin height to plate spacing ratio)/spl ges/7. In the case of H/S=4, the thermal resistance is overestimated by this method. Furthermore, nodal network analyses were carried out to estimate the thermal resistance of the plate fin including some spanwise space by specifying the cooling characteristics of the plate fin with no space around the fin. The good agreement between the predicted results and corresponding experimental data shows that this relatively simple nodal network method is quite useful to estimate the thermal resistance of plate fills including some spanwise space. >

T. Sasaki

Papers

Cooling performance of plate fins for multichip modules