FAC 2.1: Data Center Cooling Simulation Jacob Harris, Application Engineer Yue Ma, Senior Product Manager
FAC 2.1: Data Center Cooling Simulation Computational Fluid Dynamics (CFD) can be used to numerically solve very complex cooling problems in data centers. In order to continually improve in dynamically-evolving data center configurations, the CFD model needs to be at the core of any asset management solution and must be fed by real-time monitoring information. External applications are not as effective in realizing continuous improvement and validation of your cooling strategy and air handling choices. TODAY S FOCUS: Benefits of a single data model for asset management Real-time monitoring CFD simulation How CFD can benefit your bottom line
Why is CFD Increasingly Important? Energy costs are rising faster than equipment costs. CFD provides information to help users analyze their energy efficiency. By 2018, 70% of large organizations will use a program of periodic CFD analysis to continuously optimize energy costs and data center layouts, up from 25% today.** Use a program of periodic CFD analysis every six to 18 months to defer significant structural changes and capital investments (such as refurbishment or building new sites) to data centers by gaining extra cooling capacity.** Source: * http://www.vmware.com/files/pdf/virtualization-application-based-cost-model-wp-en.pdf **Optimize Your Data Center With a Periodic CFD Analysis, May 2013, Gartner, Inc. *
Why is CFD Not Widely Used? Need for DCIM-Driven CFD: Higher Process Efficiency for Faster Results 2D Layout Design Look for information Export & repair design geometry Create 3D geometry Place 1000s of servers or 100s of racks Mesh Loads BCs Solve Evaluate and Recommend The standalone process First time DCIM-Driven CFD streamlined process for subsequent data center configurations
Differences between CFD and Real-Time Environmental Monitoring CFD Predictive Simulation Tool Evaluate what-if scenarios for true cooling capacity evaluation Evaluate time-to-failure scenario Cannot be done in real-time (true engineering CFD solves for millions of Navier-Stokes equations) Can take real-time monitoring data as input and/or constraints Unlimited number of simulation temperature points (as opposed to physical sensors, which are typically limited) Provides an easy way to calculate your Rack Cooling Index (RCI ) Excellent for benchmarking Expansion recommendations Real-Time CFD This is NOT CFD, it is Real-Time Environmental Monitoring. Other than the look-alike display, it has nothing to do with CFD You can see the current state of affairs Helpful to look at trends Get alarm based on monitored temperature Limited to number of physical sensors in the room Interpolation schemes will not highlight a hot spot unless there is a sensor at the hot spot location
3D virtual model of the white room Raised floor dimensions and obstructions Perforated tile or dynamic tile location and properties Rack and IT asset inventory Average, maximum, or latest IT asset power usage and associated fan rate CRAC and CRAH operating conditions Air duct design or other air return or confinement details Temperature sensors or other monitoring data constraints you can use to simplify the CFD model Traditional Inputs to CFD
Traditional Outputs from CFD (Part 1) Simulation results can be displayed with graphical plots, charts, and reports. Some toolsets make it easy to generate images and reports to communicate the desired results to the data center operations team. The following simulation results are typically available to gather insights from a particular data center room configuration (real-time or what-if): Air velocity and streamlines for air particle tracking and animation Air and solid temperatures Volume flow rate at each tile Pressure & Turbulence data
Traditional Outputs from CFD (Part 2) Detailed temperature data for racks (maximum inlet/outlet temperatures) Return-air and supply temperatures for CRAC units Cooling provided by CRAC/CRAH units and in-row coolers
More Advanced Outputs from CFD Traditional CFD results are helpful but can be challenging to digest Cooling effectiveness of different design options can be challenging By using the same CFD technology, the detailed (not limited to sensors) rack intake temperatures can be established. Interpreting the modeled (or measured) temperatures is what most of us want: Dr. Magnus K. Herrlin came up with a dimensionless index as the vehicle for interpreting CFD results (or measurements) based on common standards. The Rack Cooling Index (RCI ) is designed to be a measure of how effectively equipment racks are cooled and maintained within industry thermal guidelines and standards. Rack Cooling Index (RCI) is a Registered Trademark and Return Temperature Index (RTI) is a Trademark of ANCIS Incorporated (www.ancis.us). All rights reserved. Used under authorization.
Benefits of CFD Reduce costly physical data center configuration changes by using flow simulation to understand performance before making the physical change. Achieve faster CFD results through a consistent environment that allows you to quickly move from asset change to simulation. Shorten modeling time for initial and subsequent analysis iterations. Gain further insight into air flow dynamics and other thermal management issues. Trade-off studies prior to the adoption of outside air economizers and optimization of the air delivery system.
Thermal Management with Real-Time Data From Temperature/Humidity Sensors Real-time Temperature Maps Monitored Temperature & Humidity Sensors TEMPERATURE SENSORS Detailed Temperature Trending
Thermal Management with CFD Best Way to Reduce Cooling Costs; A Look at the Data Center Supply Chain Extensive CFD simulation and detailed design Designed by best practice and some CFD Best practice concepts and organic growth High ROI if done right: Lifecylce- Enabled CFD with Datacenter Clarity LC is key to the solution Virtual Model of a Data Center 10W 140W+ 350W 1500W+ 4kW 30kW+ 150kW 25MW+ Energy Efficiency
Thermal Management with CFD CFD Model Set-Up and CFD Engineering Service Example of what you can easily do with CFD Initial set point temperature = 64 F (18 C) Final set point temperature = 73 F (23 C) Max air temperature = 100 F (38 C) 15% energy savings Cutting plane raised 1,500 mm from the room bottom Temperature uniformity = Higher Efficiency = Savings $$$$$
DCIM-Driven CFD
DCIM-Driven CFD Associative to asset Add/Move/Remove allow many scenarios to be investigated in a timely fashion
DCIM-Driven CFD Creating Your Baseline Virtual Room A DCIM-integrated CFD solution simplifies the CFD model creation and allows for better collaboration during expansion scenario investigation Asset Inventory Structure 1 Expansion Scenario #1: Cooling Eff. = 1.82 2 3 4 5 6 8 9 7 Expansion Scenario #2: Cooling Eff. = 1.73
Summary Cooling cost are rising Periodic CFD analysis CFD is predictive and detailed What-if scenarios Optimization Failure case analysis DCIM integration overcome CFD challenges Information gathering Geometry creation Updating to changes
THANK YOU For more information, visit usa.siemens.com/datacenters Yue Ma, Sr. Product Manager yue.x.ma@siemens.com