Operation and Control with LSM


Francis D. Reynolds

This is largely a response to William Turnbull's and Kim Goltermann's recent contributions to these debates, and to a July 26 e-mail from Turnbull to Reynolds via the Transit-Alternatives list. My thinking agrees much more closely with that of Kim, but I foresee a simpler, higher capacity, and hopefully more reliable operation and control system than either Goltermann or Turnbull have described. The optimum control system will definitely be different for different hardware systems; this article will specifically define and explain the operation and control of the Linear-Synchronous-Motor maglev dualmode system partially described in my "Platooning, Car Coupling, and LSM" item of July 24. That item should be read before this one, since this will assume an understanding of the system previously described.

This system is basically the HiLoMag proposed since 1996, but it has changed a little in some areas, and has been expanded and better defined over the years. I publicly examined suspended-car systems, and wheels instead of maglev, briefly a year or more ago, and I again prefer maglev. But I have never deviated from the belief that LSM is the most vital component to incorporate in a dualmode system, with or without maglev. The precise dead-reckoning data and the very close spacing that LSM can provide are most valuable.


I agree with the conclusion shared by many that the "brick-wall-criterion," the ability to make a safe emergency stop even if an immovable object suddenly appears on the guideway, would degrade guideway capacity too much to be practical. At first glance abandoning that criterion may seem rash, but the brick-wall criterion is seldom met on highways these days, and certainly not on the railroads, airways, or seaways. We would do our best to minimize the appearance of brick walls (better than we do in existing systems), but no transportation system has ever been totally safe under all conditions. Dualmode guideways will be no exception.

However, LSM maglev guideways would be very much safer than highways and railroads. There would be no engines or rotary motors, bearings, transmissions, wheels, tires, wheel-traction, or friction brakes to fail in the guideway mode. Almost nothing in the cars would ever wear in guideway mode. I would expect the mean time between disabling failures for these cars on the guideways to be one to two orders of magnitude greater than MTBFs for conventional self-powered wheeled vehicles. Yet we trust conventional vehicles with unknown inspection and maintenance daily by the millions. Also, LSM cars would be electromagnetically unable to travel in both directions on the same guideway lane; therefore there could be no head-on collisions. Likewise cars on the guideways could never collide with cars on the highways or railroads. We still have thousands of lethal grade crossings on our railroads, but we would never build guideway crossings at grade. And on the guideways we would eliminate the greatest source of vehicle accidents, human drivers.


I find no need for "emergency" brakes in an LSM system. In fact we need to ban them instead, since brakes that could desynchronize and decelerate individual cars on the guideways could be applied in error and cause collisions and pileups.

The only "braking" on LSM maglev guideways should be the regenerative braking automatically provided by the LSMs. Regenerative (synchronous) braking will automatically provide a retarding force to prevent the speed of any car from rising above synchronous on steep downgrades. Regenerative braking, thankfully, can never reduce the speed of individual cars below the synchronous speed—it is actually an automatic synchronous "governor," not a "brake."


Pure LSMs can't provide any thrust or negative thrust (regenerative braking) unless they are synchronized with their applied power. Synchronous motors are usually considered constant speed, but that assumes the usual constant-frequency power. They can run just as well at varying speed if they remain synchronized with input power of varying frequency.

I propose that LSM cars be accelerated and decelerated (in off-line ramps) by the application of AC or pulsed power whose frequency is ramped to produce the accelerations and decelerations desired. It is my understanding that this is not difficult to do with modern solid-state power-control electronics. In addition to frequency ramping, the use of variable pole-pitch in the physical ramps and cars may be advisable. The reader will find a discussion of this concept, and a proposed variable-pole LSM configuration in my debate submission of 12/05/00.


In the event of a power failure on an LSM guideway, or the intentional shutdown of a section of guideway carrying traffic, the cars on the disabled section would coast to a stop gradually and continue to hold their spacing by autosynchronization. Even though the line-power is off, these cars would still be electromagnetically coupled together through the guideway. Regenerative braking is a basic factor in autosynchronization.

Cars with lower drag than others would tend to slow down less rapidly than their neighbors, but regenerative braking would prevent them from doing so. As the cars slow down the rate of change of the frequency of their regenerated power would track the rate of change of their velocity. The regenerated power from low-drag cars would automatically go to cars with higher drag, and keep them from slowing down more rapidly than other cars do. The result would be the same as the maintenance of car synchronism in a mechanically coupled coasting train, but with LSM the coupling is magnetic. Both types of coupling can transmit thrust in both "tension" and "compression."

Dr. Richard Thornton and Dr Tracy Clark of MagneMotion Inc., and others have pointed out that feedback circuits will be required for stability in the use of LSM on dualmode guideways. I am a mechanical engineer, not an electrical. I believe all of the above to be true and practicable, but I lack expertise in the synchronous-electrical-machine field. Therefore if I am wrong in any of the above arguments I urge all readers who are more knowledgeable in this area to point out any errors or practical limitations.


Assume that the power in mile nine of a continuous northbound and southbound pair of guideways suddenly fails for some reason, but miles eight and less, and miles ten and on are still running. To prevent catastrophe in such a case the HiLoMag proposal incorporates unique isolating "Turnaround Loops" at regular intervals, and at the ends of every section getting its power from a different power line. This is all described in detail under "Partial Shutdown of the System" in http://faculty.washington.edu/jbs/itrans/hilo2.htm This Website also provides descriptions and illustrations of the high-speed independent-car switching and other details of this LSM maglev system concept.

In this partial-shutdown subsystem, at the instant of power failure or shutoff in one or more sections of guideway the system would instantly switch the depowered section into a long continuous loop. The traffic in that loop would circle back upon itself while coasting to an autosynchronized stop. Thus we would have gentle synchronous emergency stops without independent emergency brakes. Once stopped, the affected cars would manually drive off of the disabled guideway.

Meanwhile the cars that were heading toward the now isolated disabled section from both directions would be simultaneously switched into concentric turnaround loops so they would be headed back in the directions from which they came, still at synchronous speed. They would then be sent to detours by the computer. All of this may be difficult to envision without a picture. Have a look in the Website given above; the figure provided there will make it all clear. But one more point: Don't worry about the small but finite time and distance between the switch-setting commands in the cars and switch-setting complete. The circumference of the previously empty turnaround loops would be much greater than the switch-setting distance required, so no cars would be caught in limbo or collide.


This subject is covered in the above Website, and in my last previous debate contribution, so we won't repeat it all here. But briefly, the computers would use the known synchronism of the guideways to constantly calculate the exact positions of all cars on the guideways by dead reckoning. Needless to say, there would be redundancy and failsafe logic built into this computer system.

An entry (acceleration) ramp would not start to accelerate a car until that ramp's merge computer had identified a spot in the upcoming guideway traffic into which it intended to merge that car. As we will see under PLATOONING below, it would always place the entering car immediately adjacent (at the standard minimum clearance distance) to another car on the guideway.

All entering cars would be accelerated at a constant known rate, so the time required to achieve synchronism with the guideway would be known. The start of the acceleration of each entering car would be timed such that the car and the spot into which it is to be placed would arrive at the merging area simultaneously. This sounds like a risky operation when spelled out in this way, but that is just what all drivers do in merging with a busy high-speed highway, hopefully with room to spare. But on LSM guideways the constant speed would be known exactly, the acceleration of the entering cars would be precisely tailored, and the computers would be much more capable than human drivers. So the space required for a merge can be very greatly reduced. Since the spacing between cars would never change on the guideways, the merge clearance would become the final clearance between the cars in a platoon. So from the standpoint of platoon drag and system capacity, the less merge clearance the better. My guesstimate of "one foot clearance" may turn out to be unnecessarily conservative.

Let's add a final check on the guideway merges, to eliminate any remaining concerns. Entering cars need not be merged with the guideway traffic at the moment they reach guideway-synchronous velocity. They would be electrically switched over to actual guideway power while they are still traveling in a ramp parallel to the guideway lane. (Similar entry-position-adjustment lanes are available on many or most major highways). If needed, minor adjustments would be made in the position of the entering car by differential frequency control before the actual merge occurs.

But if for any reason a safe merge becomes impossible in the last seconds, the computer would abort the merge and the car would be decelerated in an abort ramp and returned to the streets. Merge-abort ramps are safety features that most highways lack.


Goltermann enumerated five reasons for platooning, packeting, or whatever we call it. With a simple LSM system I see only one reason: aerodynamic drag reduction. But that single advantage is more than ample reason for platooning, since it would save large amounts of energy, can be readily accomplished with LSM, and would have no disadvantages that I can visualize.

With LSM the individual cars become the controlled elements. As Goltermann pointed out, individual car control will maximize flexibility and minimize waiting times. The navigation computers wouldn't even recognize the existence of the platoons. The merging computers would gradually form the platoons online, and maintain them by patching holes left by exiting cars.

Without LSM it was argued that we would need large gaps between coupled platoons for adequate emergency braking. There will be no such need with LSM—the gaps between platoons can close as the traffic increases. From a theoretical standpoint, the fuller the guideways become the safer they will become, since there won't be enough space between cars to permit dangerous differences in velocity to develop (if synchronism should somehow be lost).

With LSM it doesn't matter how long the platoons are, except that short platoons would increase the guideway energy requirement. The length of the average platoon will be largely determined by the percentage of capacity at which a given guideway is operating at a given time.

The merging computers will be programmed to join entering cars with existing platoons, but not to wait for more than a few seconds for a platoon to arrive. Single cars introduced to the guideways become nuclei for the formation of more platoons. Should the platoons on a guideway grow so long that they all join each other, that guideway would be running at capacity. A limited number of individual cars would still be able to enter that guideway, but only into gaps made available as cars exit that guideway. With the large capacity provided by a single LSM guideway lane, most guideways won't be running anywhere near capacity for some years, except in such areas as Los Angeles. LA may have the first two-lane guideways in each direction.

All platoons will be formed online with no regard to where each car in a platoon enters the system, or where each one is going to exit. The cars in an LSM platoon will have nothing in common except physical proximity. Also, as argued in my most recent article here, the cars in a platoon need not be mechanically coupled together, even though traveling only a foot or less apart.

Goltermann correctly pointed out that in an LSM system it "would be impossible for an entering vehicle to catch up with a packet." But we won't need to. We will routinely eliminate gaps in the packets (platoons) left by demerging vehicles, not by moving vehicles on the guideways longitudinally with respect to each other, but by moving replacement vehicles into the moving gaps laterally.

Of course if there is no suitable gap in a platoon to merge a given car into, the merge computer will place that car at the front or the back of a platoon, whichever comes first. There will be no major advantages to being in the front, middle, or back of a platoon, but the end cars will have a little better view, and will probably hear a somewhat different aerodynamic noise. All cars on the guideways will travel safely, but in theory the cars in the middle of platoons will be even safer.


In order to merge a car into a gap in a platoon, the system will need to know that the gap is long enough. More than that, it shouldn't merge a car into a gap that is too long, or the remainder of the gap would increase the platoon drag unnecessarily. And the LSM in the cars will lock into step with the travelling magnetic wave only at discreet points, so we should make dualmode car lengths in increments that will fit these LSM steps. Car length will be part of the data fed to the computer as each car enters the system. I suggest a total of about four discreet lengths that all of the types of guideway vehicles would be built to.

The merging computer would check to see if there is a gap of suitable length coming up in the next few seconds. If the computer can't readily fill a gap, it will place that car at the front or the back of the next platoon instead. Each entry ramp in turn would try to fill an existing gap at each new merge. In moderate traffic no gap would go unfilled for long, but gaps for vehicles of less common length would take longer to fill.

Let's also talk about vehicle cross section. We would reduce drag the most if all of the cars in a particular platoon had the same height and width; the same cross section—and such a platoon would also look much better. So if we build all guideway vehicles to one of four different lengths, lets make the shorter two to one size of cross section, and the longer two to a larger cross section.

Assuming that most of the traffic will be private vehicles of the smaller cross section, most of the platoons would be "smooth" low-drag "trains" of constant cross section. The larger-transit and freight commercial vehicles would be placed in a fewer number of likewise constant-cross-section but fatter-appearing platoons. However, where necessary in order to reduce delays, vehicles of both cross sections could be placed in the same platoon.


Any guideway-qualified vehicle would be allowed to enter any guideway that had room for it, and it would keep its assigned spot there without challenge until it left that guideway. If it needed to transfer to another guideway its original priority would be terminated and it would acquire another reserved spot on the new guideway. Note that this is similar to what we regularly do when we travel the highways. When a particular guideway started to run at near capacity frequently, there would be pressure to build a second lane, just as there are pressures to build more highway lanes. But new guideway lanes would be required an order of magnitude less often.

I wish to emphasize that the operating characteristics of the kind of system I am describing are very similar to that of our highway system, and not at all like a railway system. But the hardware is quite different between highways and guideways, and the sources of the control intelligence are also quite different. In both systems the cars are largely independent of each other. In both there is a planned route, a starting time and place, and a final destination, but the exact time of arrival will not be known in advance with either system.

Both systems seem to fit the "managed anarchy" term suggested by Goltermann. In both cases there may be change of plans en route. Just like drivers on the highways, the guideway computers can never be sure of what will happen ahead. On the highways and the guideways certain rules are imposed for the common good, but each vehicle in both systems has extensive autonomy. The autonomy of the other cars in the system is the main reason why we won't know exactly what will be ahead until we get there. In my mind this kind of a system will be far superior, in a number of ways, to "clear-path" system concepts that have been studied. And the algorithms for managed-anarchy should be far simpler.

It strikes me that managed-anarchy transportation systems have much in common with democratic social systems, which may be why Kim chose the term. By comparison, clear-path systems are inefficient ineffective dictatorships.

With LSM, the guideways proper wouldn't be "managed," they would only need to be turned on and maintained. We turn on a synchronous conveyor belt and let it do it thing; we don't need to manage it in any way other than to put things on and take things off. The guideways would be exact automatic unthinking tools. The system management would occur in the acceptance or rejection of cars, in the ramps, in the navigation switching and merging of the cars, and in the billing.

The "anarchy" part of it would be that the computers would have no idea what customers they are going to receive, when, where they will appear, or where these customers will want to go. The arrangement of the cars on the guideways would also be anarchistic, except for the formation of platoons. The computers would satisfy customer needs individually and incrementally, just as our anarchistic highway system does.

Most of the computers would logically be based in the guideway system, but the cars would probably need simple computers. An important factor in the choice of where to locate a computer function should be; which location would minimize critical car-guideway data exchange?

Some differences between the operation of busy highways and that of busy guideways are obvious, however. And it is these differences that make us want to go to dualmode and retire the highways. Because of the limitations of human drivers the distance between cars on a highway operating at high speed is necessarily great; but when the traffic gets heavy and slows down, the human drivers become safely able to drive much closer to the car ahead. The slowing down combined with the closer spacing seems like the "Which came first, the chicken or the egg?" problem. The speed and the spacing are interdependent factors in the capacity equation for highways. But on LSM guideways, regardless of how close to capacity the guideways are running, the synchronous system will provide the same very small minimum spacing between cars, and the rated synchronous speed will not change. It is this combination of constant high speed and constant close spacing that will allow the guideways to provide consistently rapid service and very high capacity compared to highways.


Last modified: July 30, 2001