Dear TRNSYS users,
I am trying to use a Type 919 water heat pump coupled with a Type 997 horizontal ground heat pipe model to provide heating and cooling to a typical residential building (Type 56) to optimize its energy usage with GenOpt. This is the initial step toward a bigger optimization problem. I In doing so, I have some questions about how to design the optimization study to be consistent with the common design practices also some questions on details of Type 919 and Type 56.
My goal is to optimize the simple cost of the HVAC system for the house by varying the parameters below:
1. 1. Heat pump heating capacity
2. 2. Heat pump cooling capacity (assumed sensible=total)
3. 3. First stage heating set point
4. 4. First stage cooling set point
5. 5. Ground pipe length
I have fixed variables:
1. 1. 0.2 $/KW
2. 2. 20 $/meter of pipe installation and equipment cost
3. 3. 800 $/ton of heating/conditioning
4. 4. 20-year service
5. 5. Constant dead band temperature difference of 0.5 C
Questions:
1. I was wondering if you would choose similar optimization parameters? In other words, does this seem reasonable to assume these parameters and constants to optimize the problem in your opinion? Are the constant values close to reality in your opinion?
I am assuming continuous variable in GenOpt, with rather coarse step sizes at this time. Here is the list of initial and step sizes for optimization probelm in GenOpt:
Parameter {
Name = Th;
Min = 14;
Ini = 15;
Max = 21;
Step = 1;
Type = CONTINUOUS;
}
Parameter {
Name = Tc;
Min = 22;
Ini = 24;
Max = 28;
Step = 1;
Type = CONTINUOUS;
}
Parameter {
Name = L;
Min = 50;
Ini = 100;
Max = 500;
Step = 50;
Type = CONTINUOUS;
}
Parameter {
Name = CapH;
Min = 18991.08;
Ini = 18991.08;
Max = 75964.32;
Step = 6330.36;
Type = CONTINUOUS;
}
Parameter {
Name = CapC;
Min = 12660.72;
Ini = 18991.08;
Max = 75964.32;
Step = 6330.36;
Type = CONTINUOUS;
}
As you can see, I have set the step capacity values to 6330.36 Kj/hr which is equal to half a ton.
2. When a parameter is assigned as “STRING” in the deck file, the default TRNSYS units will be assumed, right?
3. I also have some technical questions about how to use Type 919 in this optimization. Mainly, I am not sure how to define the relationship between the heat pump capacities and the electric consumption power? Does a linear relationship like Powe=a*Capacity make sense? If yes, where can I find reliable data to make a reasonable assumption about the value of “a”?
I am not sure how the normalized performance curves handle the inputs? I assume that the curves account for the effects of four varying parameters at least: rated capacity, rated liquid and air flow rates and rated required power. So, would it be a safe assumption that if I can come up with a reasonable mapping scheme from rated capacity to power then all I need as input to the optimization algorithm is the fluid mass flow rate and capacity? I guess my main problem arises from not understanding how the “Normalized performance curves” work? I am not also sure if I can keep a constant liquid flow in my model and only change the capacity and power? Would this be a reasonable assumption that same rated liquid flow rates can exist for different capacity of the heat pump?
4. My other alternative to optimization of the heat pump specifications is to come up with a list of heat pumps with their input variables (capacity, power, air and liquid flow rates, etc…) from manufacturers (for which I am also soliciting for your comments on how and where to find data, easily transferable to TRNSYS performance data template) then perform a discrete optimization scheme that chooses between the available pumps! How about this?
5. Last but not least, is there a way to introduce the hourly building loads, humidity rations as well as indoor temperature values (obtained from a run of the building model w/o HVAC systems-temperature level control) to Type 919 in a separate file than the .BUI file? I am thinking of reducing the simulation time by approximating the input values to Type 919 from the previous time step throughout the year. I understand that the actual indoor air characteristics will not be modeled this way since the dynamic relationship between the indoor air and the heat pump will not be “emulated”. First of all, not sure which type to use to this simplifying assumption? Then, I was wondering how realistic the substitution of the building model with pre-imposed load, temperature and humidity values will be?
Thank you so much in advance for taking the time to read and respond.