Sunday, January 13, 2013

"End-Centric" Thinking about Microgrids

My friend Toby Considinehas an article in December's Automated Buildings Webazine that states "Buildings should be designed as microgrids in their own right, managing their own energy supply and quality."

A good point.  However, the view is limited to Commercial Buildings (no surprise there, it is "Automated Buildings" magazine after all).  I'd like to think in somewhat broader, more general, terms.  So let's consider "endpoint-centric" microgrids.

I used to refer to this as "load-centric" but I have realized that, by characterizing endpoints as loads. even the term "load-centric" is grid-centric.

More than one person, including yours truly, has argued that the "grid centric" emphasis of current thinking about Smart Grid is mistaken.  The truly "smart grid" requires "smart endpoints".  So what is required for "smart endpoints"?

One of Toby's comments in his article points up a primary need:
We know how to do all this, one building at a time. Members of the International District Energy Association (IDEA) have been doing this for decades. The integration cost is high, largely because it requires many hours of engineering time. Piecework energy integration cannot scale to continent-scale problems.
(emphasis mine)
Let me approach that second sentence first.  "Piecework energy integration cannot scale to continent-scale problems."  That's true, and that's the problem.  Piecework energy integration is precisely how our existing grid was built.  Each company built a system to serve its own service teritorry, and every line, every generator and substation, every transformer, capacitor and breaker was placed and installed as a bit of piecework engineering.  Eventually, they interconnected for reliability (more piecework integration).  The whole dogonned thing needs to be watched over 24/7, lest an unexpected perturbation cause the whole shootin' match to tear itself apart.

As Toby points out, this patchwork integration can't really handle resources at the endpoints, so distributed resources get turned off when the going gets sticky.  This is, to put it plainly, backwards.

Scalability is one reason for endpoint-centric design.  If each endpoint has microgrid capability, then the basis exists for that capability to scale upwards in a fractal manner.  While most definitions for the word "fractal" think in terms of "zooming in" to find finer and finer reflections of structures present on a larger scale, here I speak of reversing that approach.  Design each endpoint as a microgrid of intelligent endpoints.  Each microgrid functions as an intelligent endpoint in the microgrid one level above it.  Eventually, you have a "macrogrid" of scaled-up aggregated microgrids.

So, at the moment, we have what we have, a fainely-balanced patchwork quilt of grids, each with a asmall mumber of central control points.  We also have a vision for a ziggurat of microgrids, each responding to the conditions it observes at its boundary.  Here's the sticky question: How do we bootstrap from one state to the other?

One way (I think the best way) is to commoditize intelligent endpoints.  Make intelligent endpoints, way out there at the outlets, power pigtails, and wall warts, cheap and convienient.  Make them able to intercommunicate and cooperate.  Make end devices able to function in a virtual market and operate a "virtual market" in every facility.  Don't centralize the intelligence any more than necessary.

Intelligent endpoints within the facility, operating in "virtual markets" within themselves, connected to the existing grid, taking energy and information from the outside world as needs require, and passing energy and information to the outside world as needs and economy dictate.  That is, at least conceptually, scalable to larger grids of microgrids, and larger markets.

Let's start enabling intelligence at the end of the line, cheap, efficient, commodity intelligence.  Then work our way inward.  A number of organizations are working towards this.  Are you engaged in one of them?  Why not?

No comments:

Post a Comment