When I run the simulation I get this final line in the simulator output console:
"Too few parameters for subcircuit type "74hc163" (instance: xxu1)"

Is that with your models or the ones I posted?

 

Is that with your models or the ones I posted?

With the one you posted. With the model I was previously using, I get no errors reported in the console.

But there is still the problem with Q0 (and all other Q outputs) being 1/2 Vcc

 

Did you power the inverters like U2 and U3 in the first schematic? Requiring pull down at the inverter outputs already indicates a problem.

 

Did you power the inverters like U2 and U3 in the first schematic? Requiring pull down at the inverter outputs already indicates a problem.

No I didn't power them. The model automatically connects their power pins. See post #14
The pull down is necessary for the simulator to produce a trace. Something about converting from the digital domain (which can't produce a plot) to the analog domain (which can). See post #3

 

I'm beginning to think that simulating a large (100 ish) 74xx project is going to be more of a time waster than just prototyping the hardware in real life. Perhaps on a breadboard with a slow enough clock to not have a problem with stray capacitance. Do you know of any simulator that has good support for 74xx?

One thing to keep in mind is that SPICE stands for Simulation Program with Integrated Circuit Emphasis. It was intended to simulate circuits at the transistor level and model device behaviors well enough to be able to evaluate IC designs before fabricating them. It does this exceptionally well -- I've used it to simulate circuits with millions of transistors with extremely good agreement between the simulation and the actual chip after it was fabbed. But pure analog simulations are incredibly slow.

Using it to simulate PCB-level circuits, especially populated with "black-box" components, is not what it was intended for, so any abilities along those lines are adaptations ('kludges'). The ability to do an adequate simulation comes down to how good the device models are with regards to what you want to know. Unfortunately, many device models out there were made from a very narrow mindset assuming that certain conditions would apply and that only some behaviors mattered.