Outdoor temperature and Cooling power Relation


Datacenters use cooling units to take away hot air exhausted by IT devices. Cooling energy depends on IT energy in the datacenter. But, we know with observation that, outdoor temperature has an effect on cooling load too. If weather is hot, cooling demand ascends for the same IT load.

This is because, efficiency of cooling units changes according to outdoor temperature. This efficiency gain is also the reason for increasing datacenter temperatures.


  • You may want to know cooling load difference if a datacenter would be located in Erzurum (a cold city) or Istanbul (a relatively hot city).
  • You may want to employ a fogging system to cool outdoor units of cooling system and you want to decide on savings.

These questions come to answer the following question:

What is the relation between outdoor temperature and cooling load? 

We have a monitored system room. We monitor outdoor temperature, IT load and cooling load in the room hourly. We have collected  data between 02.11.2014 - 06.01.2015 (66 sample of daily average of outdoor temperature and Cooling Load.). We have seen that IT load changes (variance is 0,85 KWh and Average is 134.8 KWh) not so much daily. So we accept it as constant.

We have averaged cooling energy consumption daily. So Load data says that for the specified day, we have consumed (lets say) 92KWh (hourly) energy for cooling.

We have also averaged temperatures daily. So temperature data says that for the specified day, we have (lets say) 3 centigrade degrees temperature.

We can find the relation using regression technique in statistics.We have found the following graph:


If we talk about statistical figures; R is 0.55. This means, good relation between outdoor temperature and cooling load. R square is  0.302. This means, Only 30 percent of variance in Cooling load is explained by the outdoor temperature (Others effects can be IT load changes, stuff done in the room, etc)

Intercept: 88.666 (Y axis value when Temperature=0)
Slope=0,658

We can even estimate the formula :

Cooling Load=88.666 + 0,658*OutdoorTemperature

This means, if you increase outdoor temperature 1 centigrade degrees, cooling load will increase 0,66 KWh

I have 2 notes to be careful.
1- Here IT load was approximately 135 KWh. If load is bigger than relation may be bigger
2- We have been using DX types cooling units and no free cooling.. Free cooling can change the picture enormously..

If you have any idea on meaningful application of this data, please provide it here as a comment for other folks..





Yorumlar

Yorum Gönder

Bu blogdaki popüler yayınlar

create Virtual Machines in VMware with ansible

Hi Mainly I am a system administrator. To live freely, I work  project based and I give trainings. In my current project, I have some additional roles which include managing a vmware environment. I must create new VMs in vsphere 6.5 environment fast. I have ansible under by belt. so here how I have done this. Scenario: I have 5 VMs to create. My config files are the following; testVM.csv: this includes VM properties such as datastore, Network config etc. My each VM has 3 NICs.. If you need less, then you can trim these items in the csv file makeVM.yml: it creates VMs according to testVM.csv file. you will need Pyvmomi python library. you can install it with pip creds.txt: vcenter credentials are here hosts file: this must include only one IP. otherwise ansible tries to create a VM for each host in this file.. csv file format: VMname;cpuCount;MemorySize;diskSize;dataStoreName;PortGroupName_NIC1;IP_NIC1;Netmask_NIC1;defGW_NIC1;PortGroupName_NIC2;IP_NIC2;Netmask_NIC1;defG
Devamı

Yüksek Hizmet Sürekliliği (High Availability)

Resim
Elime Blueprints of High Availability (Evan Marcus ve Hal Stern) 2nd Ed. adlı kitap geçti. Buradan bazı kısımların özetini sizlerle de paylaşmak istedim. Ayakta kalabilirlik (Availability) bizde ne yazık ki, çoğu kurum tarafından göz ardı edilir. Halbuki küçük işletmeden, devasa holdinglere kadar her tip kurum kendi ihtiyaçlarına uygun bir HA çözümüne sahip olmalıdır. Önce temel kavramlar.. Güvenilirlik (Reliability): Bir bileşenin beklendiği şekilde çalışmasını ifade eder. Yani hata yapmamasını Ayakta Kalabilirlik (Availability): Servisin sürekliliğini fade eder. Yani bir bileşen arızalansa bile servis keintisiz sürdürülmeye devam eder. Bunun için fazladan güç kaynakları ve fanlar kullanılabilir. Veya başka bir bilgisayar, küme teknolojisi sayesinde, arızalanan bilgisayarın görevini devralabilir. Çabuk İyileşen(Resilient): Bir sorun ortaya çıktığında, normal servise kabul edilebilir sürede geri dönmeyi ifade eder. Yüksek Ayakta kalabilirlik (High Availability -HA): Bilg
Devamı