Author: Andrew Somerville, Partner, Cundall
There has been a recent resurgence of interest in dark labs within the UK. They can be key to some very niche, but important, breakthroughs in science. The fascination, however, lies in how the buildings exist and operate. As the name suggests, dark labs are facilities that are mostly automated, with no people occupying or utilising the space. Hence, they are not designed for people, but for automation.
The value of a dark lab lies in having an input at one end of the automation logistics chain, and the lab producing an output at the other end. Due to this functionality, they have been known to be beneficial in analytical medical work such as clinical pathology testing, molecular diagnostics, and routine biochemistry and haemostasis assays. Blood samples, for example, can be put on one of the chains and the dark lab can analyse the sample and produce outputs such as quantified biomarker results, full blood count data, PCR-based pathogen detection, or flagged abnormality reports for clinician review. For medical benefits, dark labs can have a great deal of value.
Regarding how the buildings operate, there are some misconceptions about energy use. Due to the nature of dark labs, they are not occupied by scientists or lab workers and are not designed for human occupation. Hence, the operational energy of these environments would not be impacted by the likes of lighting, heating or cooling. However, operational energy can still increase due to the higher energy requirements for the lab automaton itself.
Whilst dark labs not requiring humans sounds like a compelling cost and energy saver, it also means that all human activity and involvement need to be automated by technology. They need to automate a lot of human involvement within labs outside of the routines in research. For instance, in a regular lab, if something needs to be moved from point A to B, someone could simply lift it and place it down. Dark labs require automated systems that transport items like medical equipment in a structured and potentially careful manner.
To solve this issue would require designing more systems to enter the lab. These include systems like those required above that allow for the movement of items in different contexts. Given the transportation function of the machines, they may need external power sources and so the building could require charging points to keep them operational – all energy consumers. Other machinery like sample sorters and robotic arms will need to be factored into design and automation all requiring more energy to move. There is also growing interest in deploying edge data centres (on-site, small-scale compute infrastructure) and high-performance computing (HPC) adjacent to labs to directly support lab operations.
We have already seen this in genomics and bioinformatics facilities, advanced imaging and cryo-electron microscopy suites, and materials science laboratories where real-time data processing and simulation are integral to the research workflow.
AI has become widespread. One of the areas it currently thrives is in the research industry and oftentimes in laboratories. Rather than operating workloads via cloud services, many companies are choosing on-premises be it for speed, security, or cost savings. As a result, laboratories are sometimes acting akin to data centres. In practice, equipment can be located alongside wet-labs, specialised laboratories designed for experimentation. Having these computers in the lab will demand a higher power density, the need for cooling design, heat rejection systems, and more. If the laboratory needs to act like a data centre, it will have to be designed like it.
So, yes, on paper, dark labs can circumvent many of the energy demands that come from traditional buildings. However, they also come with systems and procedures that can outweigh any energy cost savings from removing people from the equation. Those who believe that human existence in labs uses more energy, have clearly not accounted for the several thousand intricacies that humans add to labs that will have to be done by bots, and cost millions!
