Is it also where we’re going?
Lionel O. Barthold, Dennis A. Woodford, and Maryam Salimi
The chaotic search for industry structure in the earliest days of electric power transmission ended by the introduction of the magnetic transformer, settling for once (if not for all) the argument of ac or dc. The efficient transformation of energy between voltage levels led to a generation and delivery pattern that lasted more than 60 years: vertically integrated supply companies, steady increases in generator size, plants sited near fuel and/or cooling water, and progressively higher ac voltage lines optimally routed (with help from eminent domain laws). That simple dream world ended in the 1960s after which:
- Industry structure changed dramatically.
- Generator size and transmission voltages stabilized.
- Load growth outpaced transmission capacity growth.
- Nuclear’s promise died while renewables emerged as economically significant.
- Watthour meters were allowed to run backward.
- Smart grids, once a concept, became a necessity.
- Microgrids have become a likely prospect.
But perhaps the most subtle change, one that would make Edison smile, has been the reemergence of dc, first in transmission, next in utilization, and ultimately in generation of electric power.
DC’s role in transmission began with a modest commercial introduction using mercury-arc technology. It now represents more than 130 major lines using rapidly advancing power electronic technology, having a combined capacity of over 100 GW and with operating voltages as high as 1,000 kV. That excludes roughly 50 back-to-back dc projects. On the drawing boards is a project that will exceed all previous dc applications combined, an 800-kV dc grid overlaying all of Europe’s 400-kV ac lines, extending from Scandinavia to North Africa with a combined capacity of up to 750 GW (see Figure 1). It will lower costs in participating countries by sharing the most efficient power production sites, exploiting load diversity, reduce spinning reserve requirements, strengthening the capability of underlying ac systems through the supply of reactive power, helping prevent area blackouts, and allowing access to renewable energy sources, such as wind to the north and solar to the south of Europe.
Europe’s move will likely be followed by a similar overlay in North America. In the United States, for example, the Midcontinent Independent System Operator (MISO) foresees one possibility for such a dc supergrid, based largely on benefits of load diversity, as shown in Figure 2.
High-level value drivers for the MISO supergrid concept are the economic exploitation of interregional load diversity an enhanced transfer capacity, frequency control, and reliability of underlying ac systems the reduced variability of wind and solar energy profiles through regional diversity the transport of higher-value energy for renewable portfolio standards requirements and energy arbitrage.
But the proposed MISO supergrid poses an unusual challenge: how to integrate existing lines into a grid format, some configured as grounded bipole configurations, others as floating monopoles. Some use line-commutated converters while newer links use voltage sourced converters (VSCs).
Meanwhile, utilities the world over look to increase the utilization of existing ac rights of way by converting selected lines to dc, realizing that dc allows continuous thermally limited conductor loading and that the new dc voltage need not conform to prescribed values but can be maximized within constraints imposed by environmental impacts and room for adequate insulation.
Conversions can sometimes double the ac rating, using just two of three conductors. A “tripole” dc system can boost that by about a third through a thermal averaging technique. A split bipole option can sometimes double the bipole rating by using both outer ac phase positions for one dc pole and the center phase position, reconductored for double ampacity, for the other.
Solar generation, while still a small share of the total, accounted for 32% of U.S. generating capacity additions in 2014. Its growth continues to accelerate and stimulate the prospect of microgrids that will permit local areas to share local generation independent of the existing ac grid. Solar panels generate dc!
While wind generation produces ac, in most cases it’s asynchronous ac since frequency is governed by the speed of blade rotation. Thus to be useful, it must first be converted to dc and then back to ac at a synchronous frequency. That arguably makes it a dc source! Wind farms are making remarkable inroads into Europe’s efforts to go green while retiring nuclear plants. In 2014, wind produced almost 10% of Europe’s electrical energy but accounted for 32% of generation additions. Wind-based projects are growing in the United States as well, and here, as in Europe, off-shore wind stations are increasingly attractive since ocean winds are stronger and more consistent and the adverse scenic and noise impact is more readily acceptable.
A consortium, the Atlantic Wind Connection, proposes to develop up to 6,000 MW of offshore wind-based power that may be transmitted to shore by dc, taking advantage of lower dc cable cost as well as its better dynamic accommodation and reactive support for the receiving systems (see Figure 3).
It won’t be long before the dc fraction of household load will match ac’s share. Modern lighting converts 60 Hz to dc. Television, computer, sound, alarm, and many control loads require dc. Add to that the prospect of the electric vehicle charging load, and dc takes command. Yet efficiencies within onboard household ac-to-dc conversion systems average as low as 65%. The dc output of a rooftop solar panel, converted to ac and then back to dc to operate a TV set, may lose half its energy in losses…thus a new interest in the delivery of dc directly to households, initially supplementing ac but possibly displacing it. Conventional wisdom, circa 1950, would have cried out on the issue of safety since dc arcs, providing no current zeros, were dangerous and hard to interrupt. Now power electronics interrupt dc faults with speeds unimaginable then and an order of magnitude faster than conventional ac circuit breakers.
Los Alamos National Laboratory examined the merits of using dc for microgrids rather than ac. The report concluded that dc microgrids may be cheaper, more efficient, and more flexible than an ac microgrid option, particularly with growth in locally based renewable generation and in vehicle charging load with its attendant energy storage system.
Electrical storage is central to all smart grid proposals, and among the options, one of several battery storage technologies is likely to be the winner. Initially considered as a point-of-load application, battery storage systems as high as 30 MW are scheduled for installation, again posing the challenge of transformation either to or from an ac system or, more efficiently, a dc system or microgrid, in the latter case calling for dc-to-dc transformation.
Enter the dc-to-dc Transformer
DC’s inroads into power generation, transmission, and consumption have been made without a dc equivalent to the ac transformer. In some industrial applications, conversion between dc voltages is commonplace but not without first converting dc to ac, transforming magnetically to a new ac voltage then converting a new ac voltage back to dc…hardly an elegant solution. U.S. patent files for direct capacitive dc-to-dc transformation would fill a small library but with language and symbolism foreign to power engineers and reflecting the fact that the realm of volts or millivolts involves rules quite different than those required for tens or thousands of volts. In that latter realm, low losses are important economically, not just a way to limit heating. Component limitations also differ between electronic and power realms as do insulation requirements and practically achievable power ratings.
The importance of an efficient and inexpensive dc-to-dc transformer became obvious as engineers in both Europe and North America came to realize the need for a very high-voltage (HV) transmission grid, electrically overlaying all existing individual HVac systems. It was apparent that grids of that nature should be dc. DC beats ac on cost per megawatt, efficiency, controllability, and reactive support for underlying ac systems. But while dc wins on those counts, it poses the challenge of controlling flow within a large grid structure as well as how to benefit from interconnection with large existing dc circuits, some of which must reverse voltage to reverse power, some which do not, some with grounded neutral, some not. DC grids will require a means of distributing flow among lines in a manner analogous to a phase-shifting transformer within ac systems. Voltage regulation too will require the equivalent of ac auto transformers. It’s not surprising that dc grid advocatesthe development of an efficient dc-to-dc transformer was set as a high priority for 2015.
To be generally useful, a dc-to-dc transformer must
- transform bilaterally
- step voltage up or down
- minimize harmonic generation in both primary and secondary systems
- require no external control signal by transforming power proportional to ∆V, just as an ac transformer does to ∆ reducing possibilities of dc system instabilities and adverse interactions
- couple HVdc systems with different grounding and different commutation systems
- require components now commonly used in HVdc systems
- act as a dc circuit breaker, isolating faults from one side to the other
- be less expensive and lighter than the original method of two VSCs in series with an ac transformer between them.
Detailed digital simulations of a multilevel modular dc-to-dc transformer (MMDCT) meeting all of the above requirements at ratings up to 1,000 MW show results identical to CIGRE dc interconnections studies in which the dc-to-dc transformer was modeled mathematically only. Simulation in other system contexts showed it to meet all of the criteria cited above.Those simulations were then extended to large wind farms wherein dc power (resulting from accommodation of asynchronous ac) remains as dc, is stepped up to a 32-kV dc collector system by MMDCTs within each nacelle, and fed to collector busses where it is stepped up further to a transmission voltage of 320 kV for transmission to shore. This eliminates the weight and space requirements of magnetic transformers and benefits from lower-cost dc cables, both from nacelles to the collector platform and from that platform to shore. Figure 4 shows two popular configurations: 1) a parallel grouping of turbines and 2) a series configuration where the dc-to-dc voltage transformation ratio is necessarily Vo/n, where Vo is the voltage output of each nacelle and n is the number of dc-to-dc transformers in series.
The importance of dc-to-dc transformation at all levels of the power supply chain is reflected in the growing number of technical papers now focused on simpler versions of traditional dc-ac-dc architecture, traditional series/parallel capacitor switching, and capacitor-based emulation of the ac autotransformer. Comprehensive simulation work shows the MMDCT option is capable of meeting all nine of the characteristics cited earlier. Details of the MMDCT was presented at the CIGRE Symposium on HVdc and Power Electronics, 21–26 September 2015 in Agra, India.
As dc’s role expands within the generation, storage, transmission, and utilization spheres, the development of an efficient, easily adapted dc-to-dc transformer has become among the industry’s highest priority development challenges. At least one relatively simple design, using building blocks and operational technology well tested in modern converter terminals, appears to meet all the requirements for such a device with rating capability needed for even the highest transmission voltages. Transformers of that nature will be important to the eventual success of dc supergrids and microgrids and the extended integration of renewable energy resources in the world’s electrical energy mix. DC-to-dc transformers are close at hand…and where the power industry started seems also to be where it’s going.