Infrastructure: a bunch of people who need the same thing (power, water, or the ability to get to work) find a large scale solution. Cloud computing is another version of this. For a nice writeup on the case for the cloud, see this post on James Hamilton’s excellent blog.
To summarize James, most standalone computer servers are way underutilized, with typical load factors of 10 – 20% (some estimates are even lower). The rest of the time, those expensive computers just sit there. Moving to the cloud ramps up utilization, bringing down fixed costs for all users. Throw in the efficiencies that come from operating at a large scale, plus a range of other benefits, and you have a pretty strong value proposition.
Taking the discussion to bricks-and-mortar infrastructure, district energy has a similar value proposition: higher utilization and better technical options with increased scale. There are other benefits, but these are two of the big ones. But, compared to district energy, cloud computing has another big thing going for it: no geographic constraints.
Any given cloud services provider can serve anyone who is online. This is a huge, diverse pool of potential customers. They are spread over different time zones, with different usage patterns. Wall Street is headed home for the day so their computer usage drops off, but meanwhile people are turning on Netflix, plus insurance companies are starting to process claims in Tokyo. This is how the cloud computing industry is pushing utilization above 60%, and driving unit costs way down.
A district energy plant can only serve buildings in the same city (or, more typically, the same neighbourhood). Pooling different types of customers (e.g. offices and residences) increases utilization because they do have different usage patterns throughout the day. But there are limits to how much diversification you can achieve when every building is in the same timezone responding to the same weather.
Here in BC’s Lower Mainland, a district heating plant would be doing very well if it achieved a 25% load factor on an annual basis. Colder climates and unusual customers like hospitals can drive that higher, but distance prevents the huge utilization wins from, for example, serving customers in opposite timezones from the same plant.
Freedom from geographic constraints has also benefited cloud computing on the supply side: plants don’t have to be near customers, so they can get built where power is cheap and/or cooling is easy. In the heating business, you need to work with whatever is available near the customer.
This isn’t to knock district energy; this comparison with the cloud applies to just about any traditional utility or infrastructure system, where there are physical limits to how far the service gets distributed. The case for district energy in dense areas, where customers are closely grouped, can be strong. But the lack of geographic constraints puts cloud computing in a different realm. No surprise that it is such a fast-growing model.
Will Cleveland
