Do trains travel at lower speed to establish time buffer for possible delays?

Amtrak does… both "kinda", and "directly".

America, having enormous land mass, has an inherent problem that would not be solved even by dedicated transcon HSR lines: It’s just too darn big.

Just take the California Zephyr from Chicago to San Francisco, not even Amtrak’s longest. It’s comparable to Barcelona to Moscow (if it went through the Alps) except the mountain running isn’t even double-track. This creates a huge "surface area" for delays to arise. And when they arise, they tend to stack.

Contrast this with a multi-segmented European rail journey of Barcelona-Marseilles, Marseilles-Paris, Paris-Mannheim, Mannheim-Berlin, Berlin-Warsaw etc. all the way to Moscow. If the Barcelona train hits an hour delay, it may stink for the traveler, but that has no effect on the Marseilles train’s timekeeping. Whereas if the Zephyr hits an hour delay still in Illinois, the knock-on effects could haunt it all the way to California.

Think of it this way: a train running 10% late is not uncommon. On a 3-hour train you can "hide that" by padding the terminal arrival by 18 minutes. But the California Zephyr is a 52 hour train, so "not uncommon" amounts to 5 hours! Too much to hide in schedule padding. Though Amtrak tries a bit. *

Add to it the low population density of the American West. With insufficient demand for dedicated passenger lines, Amtrak is a tenant on freight railroads – where it wants to go 30-50 kph faster than all the other traffic. Those lines are beyond maxed out – the freight railroads are turning away single-car traffic these days. So having a sliding "hole" in the traffic stream is increasingly challenging for the freight railroads – though they try their level best**.

In the 30-odd years I’ve been riding Amtrak, I’ve noticed a few things have been happening.

• Amtrak has lengthened their schedules over the years, adding 5-10% to run times. This allows them to flow better with freight speeds and build in "recovery time". They can tell the freight company dispatcher "hey, we’re late" and the dispatcher can push some tin around. Most Amtrak mileage is under centralized dispatching and control (CTC).
• Amtrak already had lengthy servicing stops every 1000km (600 miles) give or take for things European trains do in the terminal – refuel, provision dining cars (2-day-long train needs full and proper meal service), water the cars and empty blackwater tanks. They are scheduled for much longer than they actually take and that seems to be getting longer.
• Amtrak has added "smoking stops" which serve no operational purpose – smoking left the trains decades prior.

All of this amounts to schedule padding to build in "recovery time". The idea is that if they clip a car in Aurora, Illinois and are delayed for an hour dealing with police, they can recover the time by clipping smoking stops and hastening the servicing stop in Denver.

None of these are "deliberately running slower than track speed", but this is. Due to the senseless Chatsworth, CA collision, Congress finally put its foot down and mandated Automatic Train Stop (as part of PTC) where Amtrak or hazmat travels. This had the interesting side-effect of raising Amtrak’s legal speed from 127 kph (79 mph) to 145 kph (90 mph) on most of its route miles — notwithstanding certain tuning needed such as timing of signaled highway crossings, which are few and far between in the American west. Slow it down to 79 by the detection point and you’re fine, and passenger trains are overpowered enough to do that usefully. Amtrak has not tightened the schedule to reflect this ability; it’s a tool they keep in their pocket.

* As long-haul Amtraks near their destination, they start to overlap with short-haul passenger service. At that point, they won’t sell you a ticket (e.g. from Sacramento to San Francisco) since you have other options. As such, they have no ticketed passengers to meet, and they can leave the station BEFORE the designated time. This allows them to "pad the schedule" by stating later arrival times than intended. If they’re on-time they look like miracle workers. If they’re actually late, they are recorded as on-time or less late.

** I have never witnessed a freight dispatcher delay an Amtrak merely for convenience of freights. Much the opposite, I’ve seen downright heroism. Southern Pacific (notoriously anti-Amtrak) had 2 opposing freights meet a northbound Amtrak at a siding too small for either freight. Rather than make Amtrak sit and wait while they did their saw-by maneuver, the freights met nose to nose north of the siding. That let Amtrak slide into the siding, and the northbound freight did a 3-mile backing maneuver, the southbound freight following it, to clear the north end of the siding to let Amtrak out. The novice might say "they made Amtrak take the siding! Freight preference!" When that’s not it at all.

Trains operate at speeds mandated by the railway infrastructure administration, and reported both in the engineer’s logbook, the line signals and (where available) on-board e-signaling (ERTMS).

There is no need for a train to slow speed unless required by:

• Yellow signal (or equivalent speed reduction)
• Reduced speed zone (e.g. maintenance works)
• Upcoming switching of track (e.g. on branches to multiple destinations)

The first accommodate the presence of a train running late in front of the convoy, so that the train does not crash on the following’s tail. Trains run late mostly because they accumulate delay at stations.

Engineers are instructed to travel at maximum allowed speed as most as possible.
Source: my uncle was a train engineer

A typical example of traffic: train A is late and waiting at a stop signal for track allowance, but must wait for another train to pass in reverse direction and free the track. Train B was expected to approach the same switch, but the presence of train A triggers the red light at previous track section, which triggers as a consequence the yellow at the second track section before A.

When train A starts to move slowly and free the section, train B will be allowed to journey into the next section, etc.

As mentioned before, it’s probably more due to scheduling than due to the actual trains driving slower.

However, your question wasn’t really about the driving itself, but about the traveling. I think it’s fair to assume that that includes the wait time in the stations.

Contrary to what other answers say, I think they answer is a clear "yes" here. A practical example would be Belgium, where traveling from Antwerp to Brussels (the two biggest cities of the country) takes 7% longer in 2022 than it did in 1990. Despite the fact that the locomotives themselves obviously drive much faster.

source

I would assume that the same applies to Germany and the Deutsche Bahn.

In the UK trains are timetabled based on "Sectional Running Times" (SRTs). This is the time that a specific rolling stock type should take to travel between two locations (the Technical Running Time) and then rounded to the nearest half minute.

SRTs exist for every piece of track in PASS to PASS mode (if you’re not stopping at all), STOP – STOP mode (if you’re starting at a station and then stopping at the next, or a mix.

A full route is put together by adding up all these SRTs, then adding in the station Dwell Times (how long you need to stop at each station) and then adding on some allowances. These can be pathing allowances, for areas where you may need to wait for a junction or a platform to clear or something, or engineering allowances, for if it’s autumn and you may have poor traction, or any other technical reason.

Train operators when they bid for a route will usually want to get a bunch of allowances before the last station because it makes it easier to hit performance targets, but sometimes they will look at moving them for other reasons.

All of these factors are included and then the train is put on a graph of all the other trains and shuffled around a bit to make sure it’s not going to get in the way of all the other trains you want to run that day.

There are little bits of slack in every part of this process, but in general you get a fairly close match between the actual quickest routing time and the real world.
So the answer is… sort of yes.

Track speed limits and rolling stock are just two of the factors. I’ve seen trains catching up on delays during the course of the journey, which should be conclusive evidence that they can go faster than scheduled. But the opposite also happens.

• It doesn’t matter how fast an intercity train could go if there is a slower regional train in front of it on the same track.
• Train stations are limited in their number of platforms. When passengers are supposed to catch a connection from train X to train Y, and others are supposed to go the other way around, both trains must be in the station at the same time, not sequentially. For larger stations, the distance between the platforms affects connection time, too. So the availablity of platforms is another constraint on the schedule.
• The schedules are also written to optimize connections and to be able to be memorized. For instance, if one train goes A via B to C every two hours and another train goes A via B to D every two hours, then they may try to arrange a train going A to B every hour, roughly at the same number of minutes past the hour.

When a train has a delay, the dispatchers will have to decide if they inconvenience other trains to bring the late train back onto the schedule. Is it worth delaying a regional train with a hundred commuters for ten minutes if that means a dozen inter-state passengers don’t miss their connection and are two hours late?

I was wondering whether trains are always traveling at their highest possible speed (assuming there’s no speed limit)? Or are they not traveling at there highest speed per default to establish a time buffer for possible delays?

Most trains (rolling stock) are seldom traveling at their highest possible speed but not for the reason you mention. The main constraints on a busy mainline in Western Europe are in fact the shape and equipment of the track and the rest of the traffic.

A train that’s nominally capable of traveling at 250 or 300 km/h can only do it on very specific stretches of railway, not on the whole network. How well maintained the track is, the state of the power line, how tight curves are will determine how fast trains can go. On this map of the German rail network, you will see that the line between Munich and Berlin is a patchwork of sections with different maximum speeds. Some trains (e.g. heavy freight trains) are not capable of a very high speed but they will seldom use the fastest lines or only run during the night. Other than that, it’s the track, not the train, that’s the main constraint.

The second constraint is the traffic on the line (here is the type of plot planners use to manage it). In practice, it will be difficult to speed up without running into the train ahead of you. For that to work, you would need to waste a lot of capacity by spacing out trains, have all trains increase their speed, or delay other trains by having them wait in stations so you can overtake them. The latter does happen occasionnally but it is difficult to organise on a busy mainline.

Recently, I have also heard about some trains running under their usual speed to save energy. If a train is scheduled to run slower than allowed by the track, it could presumably speed up but traffic would still be an issue. I have actually been on trains making up some of a delay en route but that’s not common and I doubt it’s built into the schedule. The one example I have in mind was a Thalys and I suspect the high-peed lines in Belgium and the Netherlands are not very busy (LGV Nord is quite busy).

No, trains don’t deliberately run slower to give themselves a buffer.

If a buffer is desired (and it sometimes is) then it’s done by adding time to the timetable. So if a train can get to its destination at 0956 (for example) then it may be scheduled to arrive at 0958. It will drive at full speed so it arrives at 0956, and if there is no problem then it will either slow a little when close to the station (close enough so it can be sure there are no more delays) and arrive at 0958, or actually arrive at the station at 0956 and wait an extra couple of minutes.

There are speed limits on all rail lines I know about, often different between different models of trains.

If done correctly the time table will allow for the train speed limits on the given track, but sometimes trains are replaced by others that have a slower limit, sometimes there are speed limit reductions due to work or weather circumstances and so on.

In some cases there will be slack in a train time table for some delay, in other cases there will not be.

In the Netherlands we have a lot of connections where two trains arrive to the opposite sites of the same platform and allow for the passengers to just cross the platform for the change of train. In those cases many times one train has a 5 minutes stop on that station while the other train only has 2 minutes, in which case the train that has the longer stop can make up time if it arrives later. But if it comes in too late it may have a red sign as an other train needs to use (part of) the track. So making up time is not always a given.

It will also depend on whether you have an ‘every 10 minutes’ train or a ‘twice a week’ train, and on how many other trains and connections there are and how important it is for the company to make people arrive on time in the final destination or to have trains run on time. Different countries require different things from the train companies, or sometimes the same countries at different times.
One year a train company is rated for the number of trains with delays. The next year the rating system is changed and now the train company is rated on the number of passengers that reaches their final destination on time. Or either or both within a given time percentage.

Does this train traveling from Munich to Berlin operate at 250 km/h to always have the shortest travel time?

No.

Or does the train travel at 200 km/h by default and in case of a delay, the speed could be increased to make up time?

No.

In most cases, each route has its own time-table layout that is based on a complicated puzzle of engine capability, track capability, track availability, connections to other routes, maintenance, etc.

For your example, the maximum track speed between Munich and Berlin varies between 160 km/h and 300 km/h, so the train goes at different speeds along the route.

For most efficient train systems, there isn’t a whole lot of buffer in there. In some cases, they can (and do) make up delays if engine and tracks allow it, but it typically doesn’t amount to much.

One notable exception is the Boston Commuter Rail, which does not exactly qualify as "efficient". A few years ago the governor read them the riot act since they had the worst delay performance in the US (which is already a low bar to meet). In response, they padded the schedules to allow for half speed (average 30 mi/h as compared 60 mi/h). If the train runs normally, they actually have to stop for a while in the middle of the route to run down the clock. Otherwise, they show up early.