Analysis Of Thermal Characteristics Of High-density Server And Its Thermal Design

High density electronic equipments have features of high dense components, dense power distributes and high heat dissipation per unit volume. It is effective to shorten the whole R&D process if the thermal features of the products and its defects in thermal are accurately described. Therefore, the thermal numerical simulation method and thermal design are becoming one of the most important research topics during the procedure of designing electronic equipments in recent years.This paper focuses on studying IBM X3650 2U. By using ProE and thermal simulation analysis software Flotherm, the author built the 3D model of IBM X3650 2U server, solved the steady-state field and velocity field and validated simulation results through experiments. Then some factors and rules of thumb affecting the thermal features were analyzed and researched and thermal design for heatsink and the whole server were discussed. The aim of these researches was trying to summarize some methods of numerical simulation and thermal design for high density server and give some useful suggestions for the R&D of high density server. The main studies are as follows:1. 3D model of IBM X3650 2U server was built, the modeling method for the complex components was discussed and the steady-state field and velocity field distribution were solved at room temperature. Then the influence of radiance factor to the temperature field and hole-array structure to the air flow characteristics was researched respectively.2. The experiments of temperature spot test, thermal imaging temperature test and air flow velocity test were designed and implemented. By comparing to the simulation results, these results proved that the simulation method was reliable and the solution results were credible.3. By using the heat base to replace the whole heatsink, an equivalent modeling method was proposed to simplify the analytical model in system-level analysis and save calculating time. The heatsink was optimized by using Genetic Algorithm, and its results were validated by utilizing Sequential Optimization, based on which the Design of Experiments helped to improve the efficiency of heatsink's heat dissipating.4. Discussed how the air flow inside server and key points temperature were influenced by the following four conditions: with wind tunnels, without wind tunnels, only with CPU wind tunnel and only with RAM wind tunnel. The theory of Response Surface Optimization(RSO) was elaborated and PCB layout optimization for Northbridge was implemented by making up response surface in this three ways—random pick-spot, uniform pick-spot and contour lines pick-spot.