1 August, 2019
There is unrest in the data center environment. Several thousand data center infrastructure units have declared their intentions to contaminate the ambient air. This contamination movement, under the leadership of exhaust-emitting generators, has made it difficult for data center engineers to maintain clean air and order in the data center. 6SigmaDCX is returning to simulate and analyze the critical issue of contamination to assist the overwhelmed engineers…
In all seriousness, knowing the impact of external contamination is essential to ensuring data center performance and resiliency. Contamination in building complexes can severely damage the equipment used to cool and power the complex. Most data center owner-operators believe generator exhaust to be the only source of contamination, but what they don’t realize is that the recirculation of chiller unit air is also a prominent form of contamination.
While emissions from generators contain chemicals, such as Carbon Monoxide (CO) and Sulfur Dioxide (SO2), emissions from both generators and chillers are emitted at a higher than ambient temperature, leading to contamination of supply air temperatures. The contamination of supply air can cause the performance of power and cooling units to derate, resulting in lost data center efficiency.
One of the major goals of airflow management in an external domain is to mitigate contamination. Using 6SigmaRoom’s post processing capabilities, you can assess whether your facility is operating energy-efficiently or being plagued by contamination.
Test Adverse Conditions with the Data Center Digital Twin
Figure 1. Digital twin model illustrating the layout of chillers (on the top of the building) and generators (North and East sides of the building) around a data center
The building we’ll be analyzing in this model is a generic data center, lined with 1500 kW generators on the North and East sides of the building, and chiller units on the roof. The building also has four large vents on the North side, each taking in 13,750 cfm of air, for additional cooling. If these vents see contanmination over the pre-defined limit, then they will automatically close, throwing the data center into chaos. Due to high cooling requirements and the limited space of the data center, the rooftop chiller units are organized in three rows with little space between neighboring units.
For the purposes of this blog, we'll examine this conceptual model on a hot day (100F) with a strong northeast wind. Testing a variety of conditions will help the user foresee any unpredicted issues.
Note: This is a model that was built specifically for this blog to illustrate the feature. The numbers and results obtained from this model are qualitative.
Figure 2. Result tab of the Vent Property Sheet shows the contaminant concentration
By looking at the Vent’s results tab of the property sheet, we can see that the simulated concentration of CO is 0.227 PPM, much greater than the limit of 0.05 PPM. We can use the Result Cloud object to see where the contaminants are coming from. The Result Cloud, one of 6SigmaRoom’s Environment Result Plot options, adds a colored fog to the model, which can be used to analyze results above, below or in between thresholds.
By adding a Result Cloud to the model, switching our measured variable to CO and setting an upper threshold of 0.2 PPM, we will see all areas in the model with a simulated concentration greater than 0.2 PPM.
Figure 3. Result Cloud illustrates sections with a concentration greater than 0.2 PPM
The Result Cloud depicts that the exhaust flow path leads harmful carbon monoxide to the vents, and as such, adjustments to the Eastern generators on the Northern side of the building need to be considered to deter this path. The cloud also suggests that exhaust from all other generators do not pose as significant of a threat in this simulated wind direction. Their placement and exhaust should still be monitored. It is important to note that additional conditions (wind speed, wind direction, etc.) will need to be simulated in order to paint the full picture. Further, the quantitative description of contamination levels and threshold should not be regarded as a precise prediction, but as a indicator that work is required to prevent the exhausts from being re-entrained. Had the result been slightly below the threshold at the inlets, that should still be considered a risk due to uncertainty in boundary conditions and turbulence models.
Our second cause for concern is the recirculation of hot exhaust air into the rooftop units from the rooftop units themselves, or potentially even from the generators. It will be easiest to see if there is any interaction by adding a Result Plane at the same height as the chiller’s inflow vent.
Figure 4. Add a Result Plane at the chiller’s inflow vent to assess recirculation at the rooftop level
Taking a bird’s eye view of the data center building, we can see that there are elevated temperatures above 100°F, our simulated ambient air, at the inlets of the chillers. Since we don’t see any higher temperatures above the generators or from the North or East sides of the building, we can infer that the higher temperatures are most likely from chiller exhaust air being pulled back down and used for supply air.
This data center will not be able to perform optimally under these conditions. If this was a case where the data center had not been built, the engineer in charge could take this as an opportunity to redesign the layout. Potential solutions for existing sites could be relocating generators, adding or changing the ducting on the generator exhausts or the building vents to change the flow path of the contaminants, or, a most expensive option, even adding high-efficiency filters to the building vents. Any solution should be modeled and simulated in order to ensure its effectiveness.
Mitigate Data Center Contamination and Recirculation with 6SigmaRoom
6SigmaRoom provides many ways to troubleshoot contamination and recirculation issues. This is especially useful where boundary conditions are uncertain, which is particularly the case for external airflow modeling. CFD provides the ability to investigate the impact of different solutions and identify scenarios that result in much reduced risk. If you’re looking to build and examine an external model like a boss, check out 6SigmaDCX External Modeling: Key Features You Need to Be Using. If you would like to learn more about the digital twin for today’s data center, read our quick start guide to implementing the digital twin or contact us directly.
Blog written by: Max Cosculluela, Applications Engineer
Other Recent Posts
All Aboard the Digital Twin Express!
The best performing data centers will not reach 70% of their design capacity throughout their lifet…
DCD Sydney 2019, 15 Aug, Sydney AUS
Join us for DCD Sydney!DCD Sydney is a data center conference focused on helping data center, telec…