HVAC controller with predictive cost optimization
A controller for heating, ventilation, or air conditioning (HVAC) equipment including a processing circuit configured to perform a first optimization to generate a first set of control decisions for HVAC equipment including waterside HVAC equipment that consume resources from utility providers to generate a heated or chilled fluid and airside HVAC equipment that receive and use the fluid from the waterside HVAC equipment to heat or cool a supply of airflow for a building. The processing circuit is configured to perform a second optimization subject to a constraint based on a result of the first optimization to generate a second set of control decisions for the HVAC equipment and to combine the first and second sets to generate a combined set of control decisions for the HVAC equipment. The processing circuit is configured to operate the HVAC equipment in accordance with the combined set of control decisions.
1. A controller for heating, ventilation, or air conditioning (HVAC) equipment, the controller comprising a processing circuit configured to:
perform a first optimization to generate a first set of control decisions at a plurality of time steps of a first prediction horizon for HVAC equipment comprising waterside HVAC equipment that consume one or more resources from utility providers to generate a heated or chilled fluid and airside HVAC equipment that receive the heated or chilled fluid from the waterside HVAC equipment and use the heated or chilled fluid to heat or cool a supply of airflow for a building;
perform a second optimization subject to a constraint to generate a second set of control decisions for the HVAC equipment using a full model of the waterside HVAC equipment and the airside HVAC equipment to generate the second set of control decisions at a plurality of time steps within a second prediction horizon, the second prediction horizon shorter than the first prediction horizon, wherein the constraint is based on a result of the first optimization and comprises at least one of (i) a fixed value for a variable in the second optimization, (ii) an upper or lower bound for the second optimization, or (iii) a predicted state of the HVAC equipment or a predicted state of the building at one or more of the plurality of time steps within the first prediction horizon resulting from performing the first optimization;
combine the first set of control decisions and the second set of control decisions to generate a combined set of control decisions for the HVAC equipment; and
operate the HVAC equipment in accordance with the combined set of control decisions.
2. The controller of claim 1 , wherein:
the constraint is a first predicted airside demand resulting from the first optimization;
the second optimization comprises a primal waterside optimization that is performed subject to the constraint; and
the second set of control decisions are generated by performing the primal waterside optimization and override a portion of the first set of control decisions generated by solving the first optimization.
3. The controller of claim 2 , wherein performing the first optimization comprises:
decomposing a combined optimization problem for the waterside HVAC equipment and the airside HVAC equipment into a waterside dual subproblem for the waterside HVAC equipment and an airside dual subproblem for the airside HVAC equipment; and
solving both the waterside dual subproblem and the airside dual subproblem independently, wherein the first predicted airside demand results from solving the airside dual subproblem and the portion of the first set of control decisions results from solving the waterside dual subproblem.
4. The controller of claim 3 , the processing circuit further configured to:
determine an initial estimate of a multiplier for resource supply to the airside HVAC equipment, wherein the airside dual subproblem and the waterside dual subproblem are solved with respect to the initial estimate; and
update the initial estimate of the multiplier based on a difference between the first predicted airside demand resulting from solving the airside dual subproblem and a second predicted airside demand resulting from solving the waterside dual subproblem.
5. The controller of claim 1 , wherein:
the constraint is a predicted airside demand resulting from the first optimization;
the second optimization comprises a primal waterside optimization performed subject to a second constraint on the predicted airside demand; and
the second set of control decisions are for the waterside HVAC equipment and override a portion of the first set of control decisions that apply to the waterside HVAC equipment.
6. The controller of claim 5 , wherein the first optimization comprises an integrated waterside and airside optimization that generates control decisions for both the waterside HVAC equipment and the airside HVAC equipment using a thermal model of the building and a simplified resource consumption model for the waterside HVAC equipment.
7. The controller of claim 6 , the processing circuit further configured to generate the simplified resource consumption model for the waterside HVAC equipment by performing a plurality of waterside optimizations to determine amounts of resource consumption of the waterside HVAC equipment required to serve a plurality of different combinations of airside loads.
8. The controller of claim 1 , wherein performing the first optimization comprises using a simplified surrogate model of the waterside HVAC equipment and the airside HVAC equipment to generate the first set of control decisions at the plurality of time steps within the first prediction horizon and to predict the state of the HVAC equipment or the state of the building at the one or more of the plurality of time steps within the first prediction horizon, wherein the simplified surrogate model comprises one or more simplifications of the full model.
9. The controller of claim 8 , wherein the one or more simplifications comprising at least one of:
removing at least one constraint of the full model;
increasing a length of the plurality of time steps within the first prediction horizon;
combining a plurality of zones of the building into a single aggregate zone; or
removing at least one variable of the full model.
10. A method for operating heating, ventilation, or air conditioning (HVAC) equipment, the method comprising:
performing a first optimization to generate a first set of control decisions at a plurality of time steps for HVAC equipment comprising waterside HVAC equipment that consume one or more resources from utility providers to generate a heated or chilled fluid and airside HVAC equipment that receive the heated or chilled fluid from the waterside HVAC equipment and use the heated or chilled fluid to heat or cool a supply of airflow for a building;
performing a second optimization subject to a constraint to generate a second set of control decisions for the HVAC equipment, wherein:
the constraint is based on a result of the first optimization and comprises at least one of (i) a fixed value for a variable in the second optimization, (ii) an upper or lower bound for the second optimization, or (iii) a predicted state at one or more of the plurality of time steps;
the constraint comprises a predicted airside demand resulting from the first optimization;
the second optimization comprises a primal waterside optimization performed subject to a second constraint on the predicted airside demand; and
the second set of control decisions are for the waterside HVAC equipment and override a portion of the first set of control decisions that apply to the waterside HVAC equipment;
combining the first set of control decisions and the second set of control decisions to generate a combined set of control decisions for the HVAC equipment; and
operating the HVAC equipment in accordance with the combined set of control decisions.
11. The method of claim 10 , wherein:
the constraint is a first predicted airside demand resulting from solving the first optimization;
the second optimization comprises a primal waterside optimization that is performed subject to the constraint; and
the second set of control decisions are generated by performing the primal waterside optimization and override a portion of the first set of control decisions generated by solving the first optimization.
12. The method of claim 11 , wherein performing the first optimization comprises:
decomposing a combined optimization problem for the waterside HVAC equipment and the airside HVAC equipment into a waterside dual subproblem for the waterside HVAC equipment and an airside dual subproblem for the airside HVAC equipment; and
solving both the waterside dual subproblem and the airside dual subproblem independently, wherein the first predicted airside demand results from solving the airside dual subproblem and the portion of the first set of control decisions results from solving the waterside dual subproblem.
13. The method of claim 12 , further comprising:
determining an initial estimate of a multiplier for resource supply to the airside HVAC equipment, wherein the airside dual subproblem and the waterside dual subproblem are solved with respect to the initial estimate; and
updating the initial estimate of the multiplier based on a difference between the first predicted airside demand resulting from solving the airside dual subproblem and a second predicted airside demand resulting from solving the waterside dual subproblem.
14. The method of claim 11 , wherein the first optimization comprises an integrated waterside and airside optimization that generates control decisions for both the waterside HVAC equipment and the airside HVAC equipment using a thermal model of the building and a simplified resource consumption model for the waterside HVAC equipment.
15. The method of claim 14 , further comprising generating the simplified resource consumption model for the waterside HVAC equipment by performing a plurality of waterside optimizations to determine amounts of resource consumption of the waterside HVAC equipment required to serve a plurality of different combinations of airside loads.
16. The method of claim 10 , wherein:
the constraint is based on a predicted state of the HVAC equipment or a predicted state of the building at one or more of the plurality of time steps within a first prediction horizon resulting from performing the first optimization;
the second optimization comprises using a full model of the waterside HVAC equipment and the airside HVAC equipment to generate the second set of control decisions at a plurality of time steps within a second prediction horizon, the second prediction horizon shorter than the first prediction horizon, the second optimization performed subject to the constraint.
17. The method of claim 16 , wherein performing the first optimization comprises using a simplified surrogate model of the waterside HVAC equipment and the airside HVAC equipment to generate the first set of control decisions at the plurality of time steps within the first prediction horizon and to predict the state of the HVAC equipment or the state of the building at the one or more of the plurality of time steps within the first prediction horizon, wherein the simplified surrogate model comprises one or more simplifications of the full model.
18. The method of claim 16 , wherein performing the first optimization comprises using a simplified surrogate model, wherein the simplified surrogate model comprises one or more simplifications of the full model, the one or more simplifications comprising at least one of:
removing at least one constraint of the full model;
increasing a length of the plurality of time steps within the first prediction horizon;
combining a plurality of zones of the building into a single aggregate zone; or
removing at least one variable of the full model.