What could be the reason why terminal unit reduce cooling even though the zone cannot meet setpoint?

edit

This is extremely interesting from a VAV model perspective. The terminal unit does not have a cooling capacity per say, it has a modulating air flow rate. The TU model looks at it's entering air temperature (provided by the cooling coil based on the cooling coil outlet temperature set point and the coils ability to meet that set point) and then calculates an air flow rate needed to meet the zone load. This initial flow rate should exactly meet the zone load (and zone temperature set point). Then the system model looks at the air flow rate just calculated and surmises that this flow must be the flow needed for the zone and adjusts the bounds of TU's available flow (i.e., raises the min available flow or lowers the max to hone in on a solution). This happens for each zone. Then the simulation iterates again. The cooling coil is then modeled again and provides some supply air temperature. If outdoor air is involved, the supply air temperature may be raised or lowered based on available coil capacity. Now the TUs are again simulated. Except now the range of available air flow is limited by the previous attempt to restrict the air flow through each terminal unit. It is very possible that the air flow needed to meet the zone load is not available because the system model chose to restrict flow based on the results for the previous iteration. The TU limits are ultimately restricted to the point where on successive iterations the flow and supply air temperature are constant and the solution converges.

Now on to the next zone time step. If the zone temperature has risen, the zone load is larger. This time the TU thinks it needs more flow and the limits on flow converge to a different point (e.g., at a higher flow rate) and the zone temperature moves back towards the set point temperature. I've looked at this limiting point in the code and thought that there was a point where the TU is allowed to open up the previous limit but the graph shown here (good job on that by the way) seems to say there is an issue with the convergence criteria.

Is this a bad thing? Not really since VAV systems modulate flow throughout the day and the zone temperature floats around the zone temperature set point. The interesting part here is the double camel hump in zone temperature. A true VAV system would not respond that way. Could the model do a better job of hitting the set point? Maybe, but it may result in excessive iterations for the benefit of only a slightly better answer. A VAV system responds only to the thermostat temperature. If the temperature does not drift away from the set point then the TU will not change the flow rate. So there MUST be a deviation from zone temperature set point in order for a VAV system to modulate capacity ... (more)