I'm pretty pessimistic on storage these days, but I will admit that after running some back of the envelope numbers I'm actually a lot more optimistic about chemical batteries then any of the physical storage schemes.
Pumped hydro has a recharge problem - if your reservoir is much larger then your pump, you can't recharge the system in the time of cheapest energy (daytime when solar is active) before you'll be discharging again. The "roughly 4 hours" output of batteries lines up a lot better with this.
It's still too expensive, but when you plug that into the fact batteries can go in anywhere there's space, and they look a lot more attractive and definitely way faster to build. If Sodium-Ion batteries can be made to work and hit the right price point, things could change pretty dramatically though that's pinning hope on an unproven future technology.
Pumped hydro may work better with wind than with solar.
Consider the mismatch between supply and demand as a function of time; it has a Fourier transform with components at different frequencies (this is completely separate from the frequency of AC current in the system, please note). Some storage technologies are more suitable for different frequencies and different average charging times. Form's iron-air batteries, for example, would be suitable for frequencies an order of magnitude lower than Li-ion batteries. Operating a combined system with multiple storage technologies may at times involve discharging one to charge another.
Pumped hydro has a recharge problem - if your reservoir is much larger then your pump, you can't recharge the system in the time of cheapest energy (daytime when solar is active) before you'll be discharging again. The "roughly 4 hours" output of batteries lines up a lot better with this.
It's still too expensive, but when you plug that into the fact batteries can go in anywhere there's space, and they look a lot more attractive and definitely way faster to build. If Sodium-Ion batteries can be made to work and hit the right price point, things could change pretty dramatically though that's pinning hope on an unproven future technology.