What can Data Centres learn from Chernobyl?
As constant increasing demand for IT connectivity, data storage and processing continues to grow, Arup’s Kevin Burke, leader of one of our Data Centre teams, explains the importance of having a clear plan for testing and commissioning.
Over the past decade spent working on data centre projects, I find myself often drawing parallels between this field and other mission critical systems. These systems differ across industries, from navigation systems for aircraft to communication systems for first responders. As anyone designing these systems knows, uptime is vital and commissioning is crucial.
On 26 April, it was the 34th anniversary of the Chernobyl disaster. After watching the Sky/HBO series, Chernobyl, I reflected on this tragic event and the impacts it continues to have. This reminded me of the importance of considering how lessons from this catastrophe can be applied to all mission critical facilities.
When Reactor 4 was being shut down for routine maintenance, the decision was made to complete some outstanding commissioning. The commissioning involved checking if the slowing down turbine would provide sufficient power to the pumps circulating cooling water to the core until the backup diesel generators kicked in. As the power plant was operational, this changed the risk profile of this type of test.
For me, I see a parallel between this test and black building tests for data centres. These tests typically take place before data centres start operating. By isolating the mains power and confirming that the Uninterruptible Power Supply, battery systems and data hall temperatures remain within the design conditions until backup generators kick in and sync, any bugs can be ironed out. If black building tests were to take place after customers and load were already gone live, this would elevate the risk.
Data centre commissioning can only be a priority before a data centre is live. On live sites, the customer is the priority. Where possible, this testing should always be completed before going live. If any outstanding tests do need to take place on live sites, the risk profile needs to be firmly understood and processes adapted.
Clear strategies, boundaries and timescales for the testing need to be agreed between all stakeholders. In a data centre context, the client, design teams, commissioning engineers, contractors, operations and customers are all stakeholders in the commissioning process. All stakeholders need to be informed of what testing is being done, their role, the risks and the lines of communication required. If things change during the test and the required boundaries and timescales cannot be adhered to, the test should be re-arranged. It is particularly important to devise and stick to clear testing plans on projects with phased build-outs of data centres.
Another key lesson is the importance of correctly devising and following test scripts. The test loads typically reflect the design conditions the data centre power and cooling systems will be expected to support during a failure scenario. Deviating from the script has two negative impacts: (1) you are not testing what you wanted to test; (2) you run the risk of not clearly identifying the problem encountered and, as a result, it becomes more difficult to resolve. The test script is a baseline for measuring the success or failure of tests and it is a mechanism for troubleshooting failures. To truly find out how systems will perform, test scripts need to be followed.
Comprehensive testing and commissioning is clearly a vital step for all data centre projects. By collaborating with clients, Arup pro-actively supports projects through the whole life cycle. This includes planning, design, construction, commissioning, testing and handover phases, where Arup typically deploys a site-based team, followed by continuous support in operation.