The power draw went up to 2 amps, and the oil in the brushes went up in smoke.  Then it calmed down.  I let the motor run for 30 minutes at 12 volts, then I reverse the leads.
After cooling down for 3 hours, I tried it again and it wouldn't start.  I had to dump solvent alcohol down the holes onto the brushes to clean the oil residue off.  It runs reliably now.  There is no easy access to clean the brushes, so I left that issue alone for now.

Do I hook up the DCC controller??  I'm not that stupid!  I ordered qty:40 .75 Amp Poly Fuses.
A poly-fuse is like a fuse that self-heals after it cools down.  These are useful in electronics where you simply don't trust the circuit.
If something goes wrong with the motor in the future, it will stop power to the motor.

Every step I test the motor.  I also ensured it's isolated from the frame.  Installed a THIRD !!! DCC module on the shell with the lights.

Two days later, the JB Weld has cured and I can reassemble the motor.  I power tested the motor.  Then I used a small bench vise to crimp the corners.  The other flywheel twists on by hand.

Cure time for each magnet is one full day.  It stays clamped.
To take the motor apart, I removed the flywheel away from the brushes by twisting using two pairs of pliers.
Then I pried open the crimps on the housing and slid the rotor (and magnets) out of the housing.  Don't lose the plastic washer.

And for all that's Holy...  DO NOT remove those brass spring retainers.  Just leave them be.

After cleaning out the bad glue with a dremel brush and needle file, I used JB Weld to glue the magnets back onto the housing.

The squealing was definitely the rotor running against the faces of the magnets.  I surmise that by constantly injecting lots of oil, it allowed the motor to run quietly for a while.  You can see the wear on the rotor.

My latest engineering challenge comes from an upgraded Ebay purchase GP-35.  I added a DCC decoder that shorted out after running it a few minutes.
So I greased, and I oiled, and I got it to stop squealing.  I ran it on the test track while I installed missing railing and redid the lights.
Last weekend, the motor squealed again and blew another DCC decoder.
Those decoders cost more than the engine.
I see I can buy another can motor on Ebay for $20, but me being an electronics engineer, I need to do a failure analysis.

When I disassembled the motor, the magnets came out with the rotor.  They should be glued to the motor shell, but the glue job was full of voids and lack of a proper cure.
This is the first time I've seen a DC motor do this.

I just rediscovered the basic Dynamis system has the ability to write CV registers.  I was confused and thought I needed the Pro for that.
I have the Dynamis Pro on my workbench and a basic Dynamis on my little 4x8 layout loop.
It requires the Dynamis Pro to be able to read the CV values back.

It's too bad Dynamis can't be re-engineered with better workmanship.  The software is solid.  It looks like they made too many jumps that seemed wonderful that set it back.

Get rid of the IR interface and go with RJ-45 since we can buy that stuff cheap at Lowes.

I just discovered now that there is a 'Dynamis Ultima' system (12 years ago).  It runs off a self-contained USB serial port.  Maybe someday I will tear into the Pro box port and build a PC adapter.  Some of this interface information stuff might be out there.

I recently upgraded a Bachmann Plus GP-35 to DCC.
I drilled and 'self-tapped' screws into the frame halves to get my electrical pickups.
2-56 self-tapping Philips screws can be bought by the hundreds on Ebay.  It's my 'EASY' button.

On the GP-35, the horn goes as shown.  Mine was half broken off, so I glued in a piece of an Athearn F9 horn from a scrap loco.

I'd surf Ebay for pictures of your GP-30.  I had to fix my cab railings and used those pictures as a guide.  Looks like they mount on the upper sides of the cab.

I'm assuming you are running 2-rail DC or DCC.
You need to electrically isolate (plastic rail joiners) that reverse loop.
There are electronic reverse loop control modules that automatic flip polarity when a train is detected.

Or you could simply feed power to the isolated loop through a double-pull double-throw (DPDT) switch.

Have fun !!

Back in the early 1960s, MIT had a train layout where all the future computer masters of the microprocessor age got their start.
Several trains would run a schedule around the track, using passing sidings and other interlocks.
Sequenced DC operations like this could now more easily be accomplished with PLCs.  (or Arduino acting as a PLC)
As an engineering student, this is where I'd advise you to go.
I'm an electronics engineer, so I'd go a different route.  I'd program an AVR to directly give commands to the locomotives through DCC.  I've been doing this stuff with AVR processors before there was an Arduino.  I use Bascom AVR so I can't help you with the Arduino C code.

Initial cleaning involves reflowing the solder to eliminate all the dendrites already formed.  It also heats the area to remove any moisture.

A good cleaning with Solvent Alcohol and compressed air removes the flux residue.  A good scrubbing with a Dremel with rotary wire brush also cleans up the solder mask.  Additional cycles of Alcohol, Compressed Air, and microscope inspection cleans it off.

Note in the first picture how close the blue solder mask surrounds the solder pads.

I didn't remove the solder mask on this one.  After getting as much debris out of the gap as I could, I cleaned it again then baked the circuit board area with a heat gun to evaporate any oils or water under the solder mask.

Tested for 1.5 hours -- no loss of signal.

I tried the 'extra' dynamis command station on my 4x8 sheet of plywood test track and let it run for about 30 minutes.
After that time, both handsets lost signal and the trains stopped.
I swapped out the IRDA head and it worked OK for the next two hours.  This is likely why I got the command station so cheap.

The ROOT CAUSE of these Dynamis failures is not so obvious.  Yes, the Chinese work is sloppy, but also the German Engineering also has a fatal flaw that allows this sloppiness to cripple these good products.

The BLUE Plastic layer that surrounds the solder pads is the WRONG SIZE !!!!  It should allow at least 0.020 inches of space around those solder pads to keep the flux and other soldering waste out from under this 'Solder Mask'.

Instead of protecting the PCB, it funnels solder flux UNDER the solder mask.  This forms a petri-dish of Dendrite Formation.  The hand-soldering of the connector separates and damages the solder mask.

The repair is to carefully remove the solder mask from around these pads with a sharpened small jeweler's screwdriver, then clean and apply a sealing RTV over the whole set of pads to keep out moisture.

It's a tragedy that Dynamis was so discredited for reliability because some engineer goofed on the size of the Solder Mask through-hole openings in the PCB Gerber Files.

Missed one.  After reassembling and testing, this one worked but had a strange problem.  I had to press hard on the joystick right and it always jumped several locomotives each time.
Opened it back up finding I missed some dendrites shown below.  Cleaned it and it works OK.
On a different controller, I had cleaned these battery contacts.  It worked two days ago, but not today.  I successfully tinned the positive terminal, but the negative spring was too far gone.  Filing took off half the spring.
So, I covered it in solder braid, draped over the spring.  The solder braid is soldered to the PCB.

After going through all my Dynamis stuff, I thought I was seeing things.  I really have TWO dynamis control boxes??  They both work OK.  I also have two of the EZ controllers with the dial that can only talk to 9 DCC locomotives (ID numbered 1-9).

I have no clue where I picked up the second Dynamis system.  I was probably buying more hand controllers for $30 and they threw this in for free.

Shays were meant to 'in real life' operate tight switchback, poorly built, temporary logging railroads.
Some even rode on wooden rails.  A lot of those abandoned logging railroads are now Jeep/ATV trails.

In real life, they used link/pin couplers and obsolete railcar frames they stripped down to haul logs.

You will be using knuckle couplers which may uncouple in tight curves, depending on what you're coupled to.
This is the defect you will see in using your 18" radius.  You can also buy 24" radius later to fix this.