Replace cooling coil model

[AntoineGautier]

This issue is to replace the instances of Buildings.Fluid.HeatExchangers.WetCoilCounterFlow in the library and rather use Buildings.Fluid.HeatExchangers.WetCoilEffectivenessNTU.

[AntoineGautier]

Impact on HVAC energy use

Replacing the coil model in the examples from Buildings.Examples.VAVReheat has an impact lower than 5% on cooling energy, see tables below. The modeling uncertainty is also suspected to be higher with WetCoilCounterFlow as it is configured with only 4 elements (default) which yields a bias on the wet fraction computation.

branch=master
commit=3f4a4afd6579b8c451f5ac1f187235ec875867da

.. table:: Heating, cooling, fan and total site HVAC energy, and savings of guideline 36 case versus base case.

   ===================================== ===================================== ====================================== ====================================== =====
   :math:`E_{h} \quad [kWh/(m^2\,a)]`    :math:`E_{c} \quad [kWh/(m^2\,a)]`    :math:`E_{f} \quad [kWh/(m^2\,a)]`     :math:`E_{tot} \quad [kWh/(m^2\,a)]`     [%]
   ===================================== ===================================== ====================================== ====================================== =====
                                    4.81                                 15.94                                  2.307                                  23.05   
                                   4.069                                  15.0                                  1.592                                  20.66  10.4 
   ===================================== ===================================== ====================================== ====================================== =====
  
  
CPUtime, annual_base 48'
CPUtime, annual_g36 54'

branch=issue2549_newCoilModel
commit=115f736a3bd811e9db9d2690ba3a923cb8159c1f

.. table:: Heating, cooling, fan and total site HVAC energy, and savings of guideline 36 case versus base case.

   ===================================== ===================================== ====================================== ====================================== =====
   :math:`E_{h} \quad [kWh/(m^2\,a)]`    :math:`E_{c} \quad [kWh/(m^2\,a)]`    :math:`E_{f} \quad [kWh/(m^2\,a)]`     :math:`E_{tot} \quad [kWh/(m^2\,a)]`     [%]
   ===================================== ===================================== ====================================== ====================================== =====
                                   4.809                                 15.23                                  2.301                                  22.34   
                                   4.068                                 14.35                                  1.589                                  20.01  10.4 
   ===================================== ===================================== ====================================== ====================================== =====
  
  
CPUtime, annual_base 39'
CPUtime, annual_g36 44'

Note that an "apparently significant" difference may appear on transients with no CHW flow rate as the discretized coil model still transfers heat to the water volume, whereas the quasi-static epsilon-NTU model computes a zero heat flow rate for a zero mass flow rate (Q_flow = Ɛ.CMin.ΔTEnt). See for instance Buildings.ThermalZones.EnergyPlus.Examples.SmallOffice.ASHRAE2006Winter:

Impact on simulation statistics and convergence

Two models fail to simulate with JModelica:

• Buildings.Examples.VAVReheat.Guideline36 failed with RuntimeError: Process timeout.
• Buildings.Examples.VAVReheat.ASHRAE2006 failed with CVodeError: 'The right-hand side function had repeated recoverable errors. At time 75120.588517.'.

They are excluded from CI testing in this branch.

The other models do not exhibit any issues, nor significant impact on simulation statistics, see for instance Buildings.Examples.DualFanDualDuct.ClosedLoop.

Master commit dd3df91c9bf47c13721c6f529f6f15e1fe9c703e

Integration terminated successfully at T = 172800
   CPU-time for integration                  : 7.47306 seconds
   CPU-time for initialization               : 0.041198 seconds
   Number of result points                   : 587
   Number of grid points                     : 500
   Number of accepted steps                  : 600
   Number of rejected steps                  : 310
   Number of f-evaluations (dynamics)        : 9858
   Number of crossing function evaluations   : 5109
   Number of Jacobian-evaluations            : 313
   Number of model time events               : 20
   Number of state events                    : 24
   Number of step events                     : 0

issue2549_newCoilModel commit 4f1f648d4c2b68a16cc5ffd4547ab97664160266

Integration terminated successfully at T = 172800
   CPU-time for integration                  : 7.29424 seconds
   CPU-time for initialization               : 0.0377009 seconds
   Number of result points                   : 587
   Number of grid points                     : 500
   Number of accepted steps                  : 603
   Number of rejected steps                  : 308
   Number of f-evaluations (dynamics)        : 9930
   Number of crossing function evaluations   : 6591
   Number of Jacobian-evaluations            : 312
   Number of model time events               : 20
   Number of state events                    : 24
   Number of step events                     : 0

Both branches yield the same errors such as:

Error: The following error was detected at time: 68523.55069477788
  Mass fraction X[1] = -0.0144006 of substance water
of medium "Buildings.Media.Air" is not in the range 0..1
  Failed condition: noEvent(fanSupCol.port_a.Xi_outflow[1] >= -1E-05) and noEvent(fanSupCol.port_a.Xi_outflow[1] <= 1.00001)
Error: The following error was detected at time: 106327.447196189
  In ClosedLoop.heaCoi.bal1.vol:
   The heat flow rate equals 1.07088e+06 W and the mass flow rate equals 1.1441 kg/s,
   which results in a temperature difference 223.71 K > dTMax=200 K.
   This may indicate that energy is not conserved for small mass flow rates.
   The implementation may require prescribedHeatFlowRate = false.
  Failed condition: noEvent(abs(heaCoi.Q1_flow) < 836800.0*max(1.3075377777777778E-07, abs(sinHea.ports[2].m_flow)))

Impact on simulation time

There is no clear impact on simulation time, at least valid for all simulation tools, see attached files.

• Buildings.Examples.VAVReheat.Guideline36 model simulates 8% faster with Dymola but 17% slower with OCT.
• Buildings.Examples.ScalableBenchmarks.BuildingVAV.Examples.OneFloor_OneZone simulates 79% slower with Dymola but only 19% slower with OCT.
• Buildings.Examples.ScalableBenchmarks.BuildingVAV.Examples.TwoFloor_TwoZone simulates 35% slower with Dymola but only 2% slower with OCT.

branches_compare.zip