This implies that the easiest to engineer would be back to back top hats, because assuming the train makes it over the first, at the bottom of the top hat velocity is just about a constant going into the second element. It would be difficult to do one later in the course because of loss in energy and greater speed variance. Just build the second top hat smaller than the first to account for losses. In fact, the second should be easier than the first, because the first acts as a speed filter. Launch velocity is more susceptible to misfires, and differences in weight between trains. This can be adjusted by sensors to make the narrow safety window of the first top hat. If the train makes it over the first, it is going to go over the second. What happens in the unlikely event that it does not? Most coasters are susceptible to gullying, the only places a train is probably going to get back around the circuit after being stopped is at the block breaks. Anywhere else, you connect a winch, and pull the train over the next hill, not that good of idea when the next hill is another top hat. The other option is just do the same thing when a train gets stuck in a sticky point of the track in a conventional coaster. Disconnect the articulation between trains, remove the up-stops, and use a crane to lift each car off the track and place it back in the station. If a park really does not want to risk hoisting trains off the track, the second hill can be made lower to insure velocity in to the top hat, and use trims to get the train to the proper speed, or the converse and LIMs. The real question is would a second top hat really add that much value and quality to the ride? I am not sure it does. Also, continuous box beams, and support infrastructure may be more expensive than just a turnaround. And, what about forces? You are launched, your head is forced against the back rests, then you hit the 90 degree vertical that exerts a backward force on your head and neck, but it is already supported by the head rest. Cresting over the hill and heading in to a free fall naturally places your head in a foreword position, heading in to the second top hat. Then you have a greater change in angle snapping your head backward. Most people would not consider this fun, and may potentially result in some whiplash claims. So, now you are back to spacing the top hats out, where losses and speed become a real factor, and compensating for it with LIMs or whatever new systems those crazy engineers come up with becomes a cost factor. -j