Gokul Shankaran

Validation of an advanced fan model with multiple reference frame approach

Electronic enclosures commonly use fans, including axial fans, impellers and centrifugal blowers, when cooling by forced convection is required. Computational fluid dynamics (CFD) models of electronic enclosures have traditionally employed abstract fan models. These abstract fan models possess highly simplified fan geometries and numerical flow characteristics. The fan geometry is usually a rectangular or circular planar face with or without an inner concentric no-flow region representing the hub. The flow characteristics are typically summarized by a vendor supplied “pressure head-flow rate” (P-Q) curve, which, along with the continuity equation, is applied between the inlet and exhaust faces of the fan. It is generally expected that these fan curves are obtained through careful experimentation on laboratory apparatus conforming to test codes published by societies such as ASME and AMCA.

Orthotropic thermal conductivity and Joule heating effects on the temperature distribution of printed circuit boards

A printed circuit board (PCB) comprises alternating layers of dielectric material and current carrying traces and vias. As performing system-level simulations of PCBs with detailed trace and via geometries is very costly, the present approach considers the effects of the trace and via geometry in the physical model by importing ECAD data consisting of the trace and via layout of the board and determines locally varying orthotropic conductivity (kx, ky and kz ) on the printed circuit board based on the ECAD data. In addition, the present approach considers the effects of Joule heating in the current carrying traces by utilizing multiple 2-D sources where the powermap is determined by solving the governing electric field equations on the trace. In this paper, the effects of both trace layer orthotropic thermal conductivity and Joule heating are studied on a sample PCB. Comparisons are made with earlier studies and conventional models when possible. It is shown that location of the hot spots and temperature values differ substantially if different methods are used.

Using state-space models for accurate computations of transient thermal behavior of electronic packages

Under a given set of boundary conditions, thermal performance of an electronic system is generally evaluated based on its steady state response. It is a conventional practice that actual time dependent power cycles and thermal boundary conditions are time-averaged to estimate the steady state characteristics. This approach may produce accurate results provided that the time dependency of the power cycles and/or thermal boundary conditions is small. In general, time-dependent thermal analyses with actual time-dependent boundary conditions and power cycles should be performed in order to determine the steady state behavior. Whether an experimental or numerical approach is taken, accurate prediction of transient thermal behavior of complex electronic systems is time consuming. While being less overwhelming compared to the actual laboratory experiments, fully time dependent Computational Fluid Dynamics (CFD) analysis still requires large amount of CPU time. In order to prevail over this large computational cost, a number of approximate models were developed, such as Resistor-Capacitor (R-C) thermal network approaches which have produced reasonably accurate results, and therefore have been popular in determining system transient response. It is worth noting that these approaches require some rigorous curve-fitting effort followed by an optimization process and are applicable to relatively simple systems. The present study uses state-space model to determine transient and steady state thermal behavior of complex systems as accurately as possible without compromising the speed. The aforementioned technique is applied to an SOC package placed on a printed circuit board and the results are compared to those from the fully transient CFD model computations. With a dramatic reduction in the overall CPU times, the temperature histories obtained from the state-space approach agree well with the transient CFD simulations.

Accurate Prediction of Transient Thermal Behavior of Electronic Systems With State-Space Models

Accurate prediction of transient thermal behavior of electronic systems with the use of Computational Fluid Dynamics (CFD) requires large amount of CPU time. Even though steady state response of such systems is mostly of interest, transient thermal simulations with actual time dependent power cycles should be carried out to determine the steady-state behavior. Traditionally, power dissipation values are time-averaged to obtain steady-state characteristics, which may or may not be accurate. To overcome the large associated computational cost, several different approximate models were suggested and published in the literature. Among these models, Resistor-Capacitor (R-C) thermal network approaches have been popular in obtaining the transient response for the past few decades. These approaches require rigorous curve-fitting effort followed by an optimization process and are applicable to relatively simple systems. This study presents a state-space approach to determine transient and steady state behavior of electronic systems as accurately as possible without compromising the speed. This technique is applied to a sample graphics card system and the comparisons are made with the fully transient CFD model computations. It is shown that the temperature histories obtained from the state-space approach agree very well with those from the fully transient CFD simulations, where the CPU time for the former is radically (three to four orders of magnitude) smaller compared to that of the latter.Copyright

Computations with the multiple reference frame technique: Flow and temperature fields downstream of an axial fan

ABSTRACT In numerical computations, axial fans are typically abstracted as two-dimensional surfaces, and this forms the basis for the “Lumped Fan” (LF) model. The LF-model relies on experimentally derived P–Q (fan) curves which should conform to the published test codes. Despite its simplicity, the LF-model’s accuracy depends on the application and acceptable error margin. Therefore, with decreasing error margins in thermal engineering, there has been an interest in accurate fan modeling techniques such as “Multiple Reference Frame” (MRF) model. The current effort provides a two-part validation of the MRF model results for an axial fan against the relevant experiments: comparison of the (i) P–Q curves, and (ii) temperature distribution in an electronic enclosure. In the former, both fan models exhibit good agreement with the experiments, however the flow structures determined by the two models exhibit substantial differences. Given these flow structures’ strong influence on the temperature field, the MRF model demonstrates good correlation and significantly better agreement with the measured temperature distribution, compared to those of the LF-model. In light of these findings, benefits, limitations, and possible applications of both models are also discussed.

Modeling of fan failures in networking enclosures

Modeling of fan failures in networking chassis is a challenging task. There is not enough data or literature available to accurately model fan failures. This paper embarks on a study consisting of both modeling and experimental cases to investigate how to accurately model fan failures. The study will include CFD simulations in different ways to model fan failures and also real life experimental measurements to verify the simulation concepts. Recommendations will then be made about the exact and accurate ways of modeling fan failures. The study also involves cases of fan failures for both front to back airflow (Pull Systems) and back to front airflow (Push Systems)

Spatial Variation of Temperature on Printed Circuit Boards: Effects of Anisotropic Thermal Conductivity and Joule Heating

A printed circuit board (PCB) consists of multiple layers of dielectric material and current-carrying metal traces and vias. Performing system-level simulations of PCBs with detailed trace and via geometries is very costly, the present approach considers the effects of the trace and via geometry in the physical model by importing Electronic Computer Aided Design (ECAD) data comprising the trace and via layout of the board and determines spatially varying orthotropic conductivity (kx, ky, and kz) on the PCB based on the ECAD data. In addition, the present approach considers the effects of joule heating in the current-carrying traces by utilizing multiple 2-D sources where the powermap is determined by solving the governing dc electric field equations on the trace. In this paper, the effects of both the trace layer orthotropic thermal conductivity and joule heating are studied on a sample PCB. Comparisons are made with earlier studies and conventional models when possible. It is shown that the location of the hot spots and temperature values differ substantially if different methods are used.

Advances in Fan Modeling: Issues and Effects on Thermal Design of Electronics

Forced convection air-cooled electronic systems consist of fans to provide fluid flow through the enclosure. Typically axial flow fans, radial impellers, and centrifugal blowers fall into this category. In numerical computations of flow fields in electronic enclosures, axial fans have most commonly been abstracted as planar (2-D) rectangular or circular surfaces. In some cases, these abstract or lumped models may be used to mimic impellers and centrifugal blowers as well. All of these models rely on an experimentally derived “pressure head-flow rate” (P-Q) curve (also called “fan curve”). The experiments to obtain the fan curve should conform to the test codes published by ASME and/or AMCA.Convenience and accuracy of abstract fan models are dependent on the specific application/cooling method and the acceptable error margin. The latter for the thermal design of electronics has recently diminished considerably which led to the need of using more accurate and robust fan modeling techniques such as Multiple Reference Frame (MRF) model. The authors validated this method for different types of fans against relevant experimental data previously [1,2]. As a continuation of this earlier effort, an attempt is made to examine the thermal field computed by various fan modeling techniques including MRF for air-cooled enclosures in the present work. The results show that the temperature values obtained from lumped fan model and the MRF technique differ considerably.© 2012 ASME

Modeling of Fan Failures in Networking Enclosures

Modeling of fan failures in networking chassis is a challenging task. There is not enough data or literature available to accurately model fan failures. This paper embarks on a study consisting of both modeling and experimental cases to investigate how to accurately model fan failures. The study will include CFD simulations in different ways to model fan failures and also real life experimental measurements to verify the simulation concepts. Recommendations will then be made about the exact and accurate ways of modeling fan failures. The study also involves cases of fan failures for both front to back airflow (Pull Systems) and back to front airflow (Push Systems). Normally the fans have been modeled as two dimensional entities. The fan curve measured by the vendor is used in the fan during modeling. The problem that arises with this kind of a fan modeling especially during fan failures is that the three dimensional effect of the rotor and stator blades of the fan is not taken into account. In reality, the fan blades provide a big obstruction to the flow reversal that happens due to pressure imbalance during fan failures. In this paper, we start with modeling a single fan in an AMCA wind tunnel. The complete rotor and stator geometry of the fan is modeled. We run a MRF (Multiple Reference Frame) model to generate the fan curve for the fan and compare it with the experimental fan curve. After we validate the fan curve in an AMCA model for a single fan, the paper discusses three different sets of temperature and flow data: i. Temperature and flow data in a real system with four fans modeled with two dimensional fans ii. Temperature and flow data in a real system with four fans modeled with MRF fans (full 3 dimensional rotor and stator blade geometry) iii. Experimental comparisons with the simulated data. Conclusions will be drawn based on this modeling and experimental data about accurate ways of modeling fans during fan failures in real systems.

Modeling Guidelines for Large Telecom Rack CFD Models

Large electronics systems such as telecom racks and VME bus systems contain arrays of several cards where fan trays consisting of multiple fans are employed to provide forced convective cooling. Typically, flow distributors such as slotted plates are used to distribute the flow evenly through the cards. Thus, a typical telecom rack system contains a large number of parts, with a wide range of length scales. Thus, modeling such large electronic systems with all components can be challenging in terms of solution time and computational resources. In this paper, we presented two techniques to create compact board models that capture the overall physical behavior of detailed boards: 1) the Lumped Block (LB) method, which is a geometry based simplification technique, 2) the Calibrated Porous Medium (CPM) method, which is based on wind tunnel calibrations. Although different in concept, both schemes showed similar behavior when compared to the detailed card, where the maximum difference from pressure drop across the board was less than 10% and less than 30% for the thermal resistance. Moreover, both schemes resulted in mesh reduction that is greater than 83%. Despite the relatively large deviation in the thermal part, both techniques provide good choices for system level analysis where general information about the flow structure and thermal behavior of the system is required.Copyright