Never miss a story from Data Center. Subscribe for in-depth analysis and articles.
No one wants to waste unnecessary time in the model creation phase when using a modeling software. Rather than expect users to spend time trawling for published data and tediously model equipment items one by one from scratch, modeling software tends to include pre-configured items that users can simply drag and drop straight into the model.
The items’ level of detail is especially important when modeling an environment as complex as a data center, where thermal management can make or break operations. Using detailed equipment items helps to ensure the model accurately reflects airflow and thermal characteristics.
Cadence data center software, for example, contains a library of 8000+ intelligent vendor items. This library is continuously updated to keep up with modern data center design and operations. In this blog, I’ll discuss the importance of striking a balance between necessary detail and simplification when it comes to library items. I will also introduce the star rating system we use to ensure the quality of the available items.
To ensure accurate predictions, library items should be modeled to a sufficient degree of accuracy, using as much of the necessary data sheet and equipment information as possible for a close representation of the airflow and heat transfer behavior, as well as other deployment considerations, such as physical fit. While standalone equipment may not be significantly affected by oversimplification, failure to characterize the equipment can impact data center simulation results due to how equipment interacts with one another in a data center.
There are times when the practice of approximating data, such as rounding up numbers to a lower precision level, can lead to significant modeling trouble. For example, if a person using modeling software were to round up 1U = 44.45 mm to 44.5 mm, there could be issues with planning equipment spacing. Defining a library item as 44.5 mm (rather than the exact 44.45 mm) would result in the capacity planning software reporting that 42 1U-servers would not fit in a 42U rack (42*44.45=1866.9 mm vs. 42*44.5=1869 mm; 1869 mm>1866.9 mm).
For thermal simulation, oversimplifying and/or using untested approximations can miss the important interactions of airflow and heat transfer. For example:
For the sake of accuracy then, it may seem like we should only ever use detailed library items; however, there are two main issues with only including detailed items when building a thermal model. First, there are times when manufacturer data is simply limited, or we are at the conceptual design stage and so the information is simply not set yet. Second, some items, such as cabinets, have significant levels of geometric detail. Increased geometric detail often means a greater number of computational grid cells are required to capture the detail, which can lead to longer solve times. The best solution is to strike a balance between detailed items that ensure accurate predictions and simplified items that ensure reasonable solve times and the necessary level of granularity required for the engineering decisions being sought. That’s why Cadence data center software provides users with the option to model the geometry of an item in detail or simplified. In conceptual design, capturing overall cooling performance may be all that is needed. This is because when specific equipment data is known, details will likely change resulting in a consequent change in detailed behavior.
We have created a large collection of library items specific to data centers. Built by our in-house engineers or configured by vendors, these items are constructed with manufacturer data and on-site measurement data, and rated on the level of detail using our star rating system. This star rating system offers visibility on the level of detail each item is designed for, ranging from 0-5, with 0 containing no information and 5 being comprehensively modeled.
To further streamline the design process, our library items can be dragged and dropped into place in the model as needed and are easily editable. The items can also be mapped from imported models, which means less time can be spent on the model creation stage and instead on the thermal analysis stage. The below video demonstrates how easy it is to use our library of intelligent objects.
The right simulation software can make it easier to build and interpret model data, so teams can focus on making the best decision for their data center’s success. We regularly update our library of smart items to ensure the model build process is as easy and efficient as possible.
Learn more about how Cadence data center software provides the performance visibility data center teams need to optimize data center performance.