Transient Thermal Simulation Model for Cyber-physical Simulation of a Computing Infrastructure Environment

Sandeep Gupta (Inventor)

Research output: Patent

Abstract

Data centers annually consume $7.4 billion dollars of electricity. The majority of the energy is used for support infrastructure such as air conditioning rather than electricity being used by computers. Computer programs that predict temperatures in key locations of data centers can be used to make decisions regarding the proper server to assign to incoming work. Lowering demand on a server reduces processor heating and shuts off cooling fans, which saves electricity, therefore improving the data centers operating bottom line. Thermal modeling software currently available on the market can only predict steady state temperatures. This ignores the oscillating behavior of cooling where the chiller removes heat at a different rate than processors generate heat. These programs also ignore thermal capacitance of solid material, which understates the amount of time equipment is overheated, thus shortening the life of the equipment. Researchers at Arizona State University have developed a new simulation model for data centers. This model accurately predicts temporary temperature variations that steady state models cannot. Unlike simulations based on computational fluid dynamics, which take hours to generate models, this program creates simulations in a few seconds. The model can also predict event triggers such as CPU throttling that can have a lasting effect on computational performance. This innovation was designed to be used in a workload placement algorithm, which can schedule the workload. This allows improved energy efficiency of the data canter, reduces repairs to computers, and extends the life of the equipment. Potential Applications Workload scheduling Thermal modeling Benefits and Advantages Lower Costs Reduces the amount of electricity used to cool computing equipment and extends equipment lifetime. More Power Allows computer processors to operate more efficiently. Retrofit Can be used with existing workload scheduling software. Download Original PDF For more information about the inventor(s) and their research, please see Dr. Sandeep Gupta's directory webpage Dr. Georgios Varsamopoulos's directory webpage
Original languageEnglish (US)
StatePublished - Mar 8 2013

Fingerprint

Electricity
Servers
Scheduling
Cooling
Air conditioning
Temperature
Fans
Energy efficiency
Computer program listings
Computational fluid dynamics
Repair
Capacitance
Innovation
Hot Temperature
Heating
Costs

Cite this

@misc{46865f474f4340fb802607ad4ce2a7f1,
title = "Transient Thermal Simulation Model for Cyber-physical Simulation of a Computing Infrastructure Environment",
abstract = "Data centers annually consume $7.4 billion dollars of electricity. The majority of the energy is used for support infrastructure such as air conditioning rather than electricity being used by computers. Computer programs that predict temperatures in key locations of data centers can be used to make decisions regarding the proper server to assign to incoming work. Lowering demand on a server reduces processor heating and shuts off cooling fans, which saves electricity, therefore improving the data centers operating bottom line. Thermal modeling software currently available on the market can only predict steady state temperatures. This ignores the oscillating behavior of cooling where the chiller removes heat at a different rate than processors generate heat. These programs also ignore thermal capacitance of solid material, which understates the amount of time equipment is overheated, thus shortening the life of the equipment. Researchers at Arizona State University have developed a new simulation model for data centers. This model accurately predicts temporary temperature variations that steady state models cannot. Unlike simulations based on computational fluid dynamics, which take hours to generate models, this program creates simulations in a few seconds. The model can also predict event triggers such as CPU throttling that can have a lasting effect on computational performance. This innovation was designed to be used in a workload placement algorithm, which can schedule the workload. This allows improved energy efficiency of the data canter, reduces repairs to computers, and extends the life of the equipment. Potential Applications Workload scheduling Thermal modeling Benefits and Advantages Lower Costs Reduces the amount of electricity used to cool computing equipment and extends equipment lifetime. More Power Allows computer processors to operate more efficiently. Retrofit Can be used with existing workload scheduling software. Download Original PDF For more information about the inventor(s) and their research, please see Dr. Sandeep Gupta's directory webpage Dr. Georgios Varsamopoulos's directory webpage",
author = "Sandeep Gupta",
year = "2013",
month = "3",
day = "8",
language = "English (US)",
type = "Patent",

}

TY - PAT

T1 - Transient Thermal Simulation Model for Cyber-physical Simulation of a Computing Infrastructure Environment

AU - Gupta, Sandeep

PY - 2013/3/8

Y1 - 2013/3/8

N2 - Data centers annually consume $7.4 billion dollars of electricity. The majority of the energy is used for support infrastructure such as air conditioning rather than electricity being used by computers. Computer programs that predict temperatures in key locations of data centers can be used to make decisions regarding the proper server to assign to incoming work. Lowering demand on a server reduces processor heating and shuts off cooling fans, which saves electricity, therefore improving the data centers operating bottom line. Thermal modeling software currently available on the market can only predict steady state temperatures. This ignores the oscillating behavior of cooling where the chiller removes heat at a different rate than processors generate heat. These programs also ignore thermal capacitance of solid material, which understates the amount of time equipment is overheated, thus shortening the life of the equipment. Researchers at Arizona State University have developed a new simulation model for data centers. This model accurately predicts temporary temperature variations that steady state models cannot. Unlike simulations based on computational fluid dynamics, which take hours to generate models, this program creates simulations in a few seconds. The model can also predict event triggers such as CPU throttling that can have a lasting effect on computational performance. This innovation was designed to be used in a workload placement algorithm, which can schedule the workload. This allows improved energy efficiency of the data canter, reduces repairs to computers, and extends the life of the equipment. Potential Applications Workload scheduling Thermal modeling Benefits and Advantages Lower Costs Reduces the amount of electricity used to cool computing equipment and extends equipment lifetime. More Power Allows computer processors to operate more efficiently. Retrofit Can be used with existing workload scheduling software. Download Original PDF For more information about the inventor(s) and their research, please see Dr. Sandeep Gupta's directory webpage Dr. Georgios Varsamopoulos's directory webpage

AB - Data centers annually consume $7.4 billion dollars of electricity. The majority of the energy is used for support infrastructure such as air conditioning rather than electricity being used by computers. Computer programs that predict temperatures in key locations of data centers can be used to make decisions regarding the proper server to assign to incoming work. Lowering demand on a server reduces processor heating and shuts off cooling fans, which saves electricity, therefore improving the data centers operating bottom line. Thermal modeling software currently available on the market can only predict steady state temperatures. This ignores the oscillating behavior of cooling where the chiller removes heat at a different rate than processors generate heat. These programs also ignore thermal capacitance of solid material, which understates the amount of time equipment is overheated, thus shortening the life of the equipment. Researchers at Arizona State University have developed a new simulation model for data centers. This model accurately predicts temporary temperature variations that steady state models cannot. Unlike simulations based on computational fluid dynamics, which take hours to generate models, this program creates simulations in a few seconds. The model can also predict event triggers such as CPU throttling that can have a lasting effect on computational performance. This innovation was designed to be used in a workload placement algorithm, which can schedule the workload. This allows improved energy efficiency of the data canter, reduces repairs to computers, and extends the life of the equipment. Potential Applications Workload scheduling Thermal modeling Benefits and Advantages Lower Costs Reduces the amount of electricity used to cool computing equipment and extends equipment lifetime. More Power Allows computer processors to operate more efficiently. Retrofit Can be used with existing workload scheduling software. Download Original PDF For more information about the inventor(s) and their research, please see Dr. Sandeep Gupta's directory webpage Dr. Georgios Varsamopoulos's directory webpage

M3 - Patent

ER -