Users of a ‘microgrid’ – a system that supplies and manages power for a physical entity such as a store or factory – would be better off if most of the power used was direct current (DC) electricity generated locally, argues Paul Savage.

When it comes to Direct Current (DC), confusion reigns about where to draw the line around the concept of a microgrid. For utilities, the idea is a subset in the system, encompassing many customers. For customers, it is often all the equipment on their side of the meter. For regulators, it is an issue that needs definition due to the pressing need to get the most out of DC for the benefit of the entire system.

A microgrid allows the Hempstead Town Hall in New York to use solar PV at greater than 98% efficiency
Click here to enlarge image

My company, Nextek Power Systems, has pioneered the use of DC power systems for over 10 years, originally as an efficiency booster for the entity that pays the bill. What we have discovered is that the benefit of adopting DC power networks extends beyond higher integrated efficiency for DC power generators that support DC loads. Control of these systems can be implemented more cost-effectively, as can storage, but the topic of this article is the benefit to the whole grid system and all its stakeholders, either directly or indirectly.

By using the managed output of distributed generation assets where, when, and how the power is created, we can provide greater efficiency and less reliance on the utility grid. Those who promote transforming the power gird into a two-way street will not capture primary benefits that DG should inherently bring: decreasing congestion on the grid and de-emphasizing the grid as the only source of power. The utility grid was never designed to accept power back from locally generated sources. Nor, by definition, can it sustain the customer’s needs when it fails. Let’s take the opportunity to fill these gaps with a local architectural solution that is under each customer’s control.

The utility grid, the master distributor of AC power, is far from a local user’s problems and needs. Local, predominately DC networks are called microgrids for good reason: they are very tiny compared to the national grid, and they give customers control over their power supply as never before. Microgrids allow customers to deliver power to their critical DC loads independent of the condition of the grid at-large.

Let’s see how various definitions of a microgrid have failed to capture a system that can be universally useful to every customer that deploys one.

Historically, proponents of distributed generation have designed in the AC domain by habit, which has proved to be inappropriate for what is becoming the typical microgrid load. AC began and exists as a better method of distributing power over long distances, but DC power is both the currency of renewable generation as well as the life blood of all electronic devices. How can this reality be accommodated, and how can both forms of electricity – AC and DC – be best exploited by end-users?


Definitions are often arbitrary driven by self-interest. For instance, DTE Energy, the seventh largest utility in the US, has defined microgrids as ‘small-scale energy supply and delivery systems that generate power and produce thermal energy on-site or adjacent to the multiple customers and facilities they serve.’ This is hardly a standard definition.

The combiner box where the PV panels get connected in series to build up the desired voltage
Click here to enlarge image

The EU’s European Research Area, regarding distributed generation, defines a microgrid but does not mention any specific application of power generated, nor auxiliary characteristics such as thermal output in its description: ‘Microgrids comprise low-voltage distribution systems with distributed energy sources, storage devices and controllable loads, operated and connected to the main power network or islanded, in a controlled, coordinated way.’

The closest definition we’ve found that approaches the Nextek (our) concept of a microgrid comes from the Oak Ridge National Lab (ORNL): ‘A microgrid is defined as an aggregation of electrical loads and generation. The generators in the microgrid may be microturbines, fuel cells, reciprocating engines, or any of a number of alternate power sources. A microgrid may take the form of a shopping center, industrial park or college campus. To the utility, a microgrid is an electrical load that can be controlled in magnitude.’ The load could be constant, increased at night when electricity is cheaper, or held at zero during times of system stress.

These definitions do not characterize the kind of power network to be used. And they make no reference to exporting any locally generated power back to the utility grid. Therefore, one can conclude that for a microgrid to be a microgrid, it does not have to supply power outside its managed domain. If this happens to occur profitably, it is a secondary effect only.

Nextek has chosen to more carefully and holistically define a microgrid as follows: The microgrid is an independent system that supplies and manages power for a defined physical entity – such as a store or factory – that will accept generated power from any and all kinds of power-generating sources. In most cases, the common source of power outside the microgrid will be AC from the utility grid, as it should be. The microgrid will, as required, generate and store power locally from a variety of chosen technologies that are driven by each client’s specific requirements. Excess generated power will, preferably, be stored serving client back-up or peak-mitigation needs. Most locally available power will be converted to or generated as DC because DC is the most common currency required by today’s electronic devices. Meanwhile, AC power supply will not be eliminated entirely.


Except for rotating generators delivering 50 or 60 Hz or a multiple thereof, all generation is intrinsically DC. Indeed, even high-frequency AC generators must rectify (or ‘make into DC’) their electrically noisy output before ‘good’ AC can be made – that which is suitable for grid delivery. Therefore, if DC is being made at a specifically defined site such an office building, that power, without further modification, should be used to supply directly all DC devices which include everything electronic. These include all kinds of modern lighting systems, every device with semi-conductors (computers), motor drives that modulate standard AC motors, every sensor and wireless device.

Alternatively, if the supply is greater than the demand for the power, it should be stored as DC for later use. External factors may also drive this choice, and so excess power can be exported as AC as conventional practice, if local rules make that financially more attractive.

Distributed generation’s promise is best realized in the DC domain because the customer can avoid the losses incurred by conventional systems using inverters. In conventional practice, the DC power generated on-site is inverted to AC – the denomination of the grid – only to be immediately transformed back to DC (rectification) for use in the actual loads the customer needs to operate. Consumers can raise their integrated DG efficiency by 10% or greater by at least partially adopting a DC network or microgrid. This is a result of choosing which loads will benefit from the DC, which is one of the sources that feed the microgrid. This 10% minimum is made up of 5% avoided inversion losses (going from the native DC to the AC denomination of the grid), and at least 5% rectification losses at the load (this is usually very much higher and is regularly over 20%).

The Electric Power Research Group (EPRI) has written a white paper that encompasses this topic and related issues. It concludes that DC data centers are 20% more efficient than best-in-class AC ones, and 30% better than AC systems more commonly found in the field.

But this efficiency improvement (which can be as high as 25%) has a broader benefit that extends beyond the individual customer. The grid operator benefits from not having to deliver those kilowatt hours over the grid, which was originally conceived of as a one-way street to the consumer. All the other users benefit from the increased efficiency of distributed generation because the transmission and distribution (T&D) losses imposed on the distributed generation are eliminated. This is to the benefit of all (those who use it and all those who help pay for it). This T&D loss is estimated to be over 7% in the US and the UK during the past decade, according to the US Climate Change Technology Program and the UK’s Powerwatch. The US Department of Energy’s Ann-Marie Borbely, in her book on distributed generation, estimates that the losses will fall in a range of 7%-11% in most service areas.

Therefore, a fully integrated picture of distributed generation power networks to be considered by industry, governments and other stakeholders must include the avoided conversion losses plus the avoided T&D losses, which together falls between 17% and 35% – a range that demands attention.


It is often said that if only storage could be integrated into the grid infrastructure, many of our problems could be solved. This is, of course, impossible in the AC infrastructure because AC is dynamic – only DC can be stored, either chemically (e.g. a battery), physically (e.g. a capacitor), and/or kinetically (e.g. heat, weight, etc.). A primary benefit of adopting DC power networks close to the customer is the seamless integration of battery storage for use in demand side management. A DC system increases the value of this strategy by at least 10% versus using batteries to serve AC loads during peak demands. In actual practice the number usually exceeds 20%. This is due to the low efficiency of most AC-to-DC power supplies that come with the vast majority of appliances we buy. This is driven by price. We want the computer, DVD player, printer and other electrical devices available at the lowest price possible. We fill offices with these devices and treat them as commodities.

Microgrids are ideal for building-integrated power sources which serve on-site loads
Click here to enlarge image

Such is the supremacy of batteries to this date that they continue to be the only distributed solution that is well deployed. Indeed, two US companies (Gridpoint and Gaia Power Technologies) are selling battery-based systems that are supporting AC loads, betting on the customer’s distaste for gensets and utilities and the dire need for peak mitigation during periods of high demand. This phenomenon comes on the heels of the success of two other companies (EnerNOC and Comverge) that broker the option between utilities and their customers to either turn down or turn off their loads according to a pre-arranged agreement.

Both business plans (the battery integrators and implementers of demand-side management) would be better served by DC power systems. Indeed, where these systems are also fed by rectified grid power, they also offer the customer the benefit of operating during a power outage. Further, unlike inverter-based systems, they operate without the worry or technical need for an interconnection agreement and interface, because they do not need to make AC for export. There are dark scenarios associated with improperly managed AC export such as the danger to linesmen trying to fix grid problems.


This idea for microgrids is really about buildings and groups of buildings acting in consort, and how to make that performance more efficient. There are many near-zero and positive energy building concepts around the world that have different approaches, but each and everyone of them can benefit from the introduction of a DC power network. This feature would also make renewable energy ready, as well as simplifying integration requirements.

Power sources that are truly integrated into buildings – such as building-integrated photovoltaics (BIPV) or integrated small wind turbines – are obvious candidates, but so are all electrical power generators that can be configured for DC output.

An intriguing dimension to this updated architecture regards conventional natural gas retailers. It is a rather unexpectedly belated route back into the lighting business, consisting of gas-powered high-speed DC generators feeding an electronically ballasted building lighting system. Such systems are not only very efficient, but are also grid-independent (i.e. un-interruptible).

Lastly, the idea of establishing a DC power network which is connected, but not inter-connected to the larger utility grid, also raises the promise of higher-efficiency AC systems. This is due to the possibility of introducing a rectifier of high-efficiency between the grid and the DC power network, in which the rectifier, and any subsequent DC-to-DC conversion, loses less power than the efficiency of the rectifier found in the load supported.

Nextek has ample evidence supporting this from its experience with lighting systems. For example, 400-Watt HID fixtures, where lamp drivers are known to be very inefficient, gain significant improvement by omitting the drivers and feeding the ballast 380-Volt DC directly. This amounts to a 5% efficiency improvement for the lighting system formerly dependent on AC, and over a 15% improvement when driven by locally generated DC. Using on-site generated direct current to support intrinsically DC loads just makes more sense.


The idea of exporting power to the grid appeals to everyone on a gut level, but it turns out it’s not for good reasons. While it may be a compelling mental image, both digital (countdown to money in-flows) and analogue (the flat meter disk spinning from right to left), in fact the economics support turning your meter more slowly by adopting more efficient power-consuming habits, and then addressing the resulting considered load. It’s a fool’s errand to strap solar panels on your roof if you haven’t got efficient appliances inside the building.

Indeed, the electrical grid and the loads it serves are a system designed to flow in one direction. And if distributed generation systems were engineered for top efficiency and a one-way connection to the grid, the grid would be more robust as a whole. The idea of having a little ‘money maker’ in the basement, making power to sell to your neighbours, has no barrier to entry, but it does, however, have several important factors running against it.

First, the legitimate needs of the power companies and the grid system operator. How will the cost of the T&D wires be equitably shared by all users? This has already been a notable barrier to large distributed generation installations and the establishment of ‘net-metering’ rules that are supposedly required for Renewable Portfolio Standards (RPS) for US state jurisdictions. Distributed generation should be seen by all stakeholders as the efficiency measure it is, and not reliant on net-metering rules that promote inefficiencies.

Secondly, looking at the loss in inverting the power generated, and conditioning it to get it in phase with the utility power. Power generated on-site and fed into a DC microgrid avoids conversion losses. This is at least 5% or better for the small kinds of systems we are considering here. There are experts who will complain that the loss in the T&D itself should not be borne by the utility which pays the producer in the net-metering scenario. Does it make sense for the utility to pay for power it cannot dispatch, or should we give customers a portal to serve their own loads, in addition to the utility-provided electricity?

It is hard to argue against the success of the German feed-in tariff. But we can’t use it in its present form because it doesn’t explicitly measure DC kilowatt hours used on-site. This is an ambiguity we intend to test in Bavaria, but given the efficiency at stake, this should be the next logical evolution of the feed-in tariff: turn it into the ‘use-it but meter-it’ tariff and eliminate the necessary middleman.

We see in this proposed architecture a parallel to the sun’s own delivery mechanism, the ultimate distributed resource. If quality is the objective and unnecessary losses are to be avoided, it is clear that converting, and then shipping, newly released electrons into the grid when they could be used as-is nearby is akin to continuing to use inefficient and wasteful incandescent light bulbs because that’s what your parents did.

Electronics that consume DC power as opposed to AC power have an increasing role in every level of our society. Renewable energy which produces DC has captured not just the popular imagination, but also the fastest growth rate amongst distributed generation technologies. And in that, we can find both a DC standard to enthusiastically promote and many new opportunities for improved efficiencies in the use of the distributed generation – the microgrid concept.

Paul Savage is CEO of Nextek Power Systems, Inc., New York, US.