Typically, piston pumps are positive displacement devices. For an ideal pump, for a given pressure differential, they move the same amount of fluid per cycle no matter how fast the pump turns. Also, the average amount of torque needed to turn the pump's crankshaft doesn't change with the pump's speed.
This
design of piston pump is not a positive displacement one. When the crankshaft is turned very slowly, no fluid is pumped, and the torque required to turn the crankshaft is zero, or close to zero. At higher speeds, the amount of torque needed to maintain a constant speed goes up, and the amount of fluid moved per rotation goes up.
At the center of the design is a 4 opening manifold, shaped like a "+", a plus sign. One of the openings of the manifold has a check valve which will let fluid into the manifold from the (low pressure) fluid supply. In the opening opposite, there is a check valve which will let fluid out of the manifold, into the (high pressure) outlet. The third opening leads to a cylinder with a piston, which in turn is driven by a crankshaft, which is driven by an appropriate motor.
If the fourth opening of the manifold were capped off (which it won't be) then we'd have a fairly conventional positive displacement piston pump.
Instead of being capped off, the fourth opening leads to a hydraulic accumulator. This accumulator should be a modified raised weight design. Specifically, a very heavy mass pushes on a piston in a cylinder, with the fluid being pumped in the region below the piston. The variation from the norm is that the weight is also partly supported by a compression spring, to prevent the weight from bottoming out.
Now, it should be fairly obvious why, at low speeds, very little torque is required to turn the crankshaft of the pump -- doing so simply shifts fluid back and forth between the accumulator and the pump's main cylinder.
It *might* be less obvious why the pump will be able to move fluid when turned at high speeds.
When the crankshaft of the pump is moving at high speed, the mass of the accumulator acquires a high speed sinusoidal motion, which of course also requires a high speed sinusoidal acceleration. When the accumulator's weight is at it's lowest point, it is experiencing it's highest upward acceleration... this acceleration can only be partly supplied by the spring; the remainder must be supplied by the fluid being pumped.
At this time, the pump's main cylinder is at it's maximum displacement -- due to the angle of the crankshaft, the piston can't move any further. Thus the momentum of the mass of the hydraulic accumulator forces fluid out the check valve into the pump's outlet, until enough kinetic energy has been converted into hydraulic energy for the piston to change direction.
Something similar happens as the mass of the accumulator reaches it's highest position -- but producing low pressure, which sucks water in from the inlet.
This pump can be best understood by comparing it to George Constantinesco's torque converter, but with the crankshaft and connecting rod being replaced by the crankshaft, connecting rod and piston; with the power splitting lever being replaced by the manifold; with the overrunning clutch being replaced by the check valves of the manifold; with the inertial mass of the pendulum being replaced by the mass of the hydraulic accumulator.