energymusings-blog

[energymusings-blog]

A common issue for air handling units (AHUs) is the failure of the air-side economizer to perform as intended. Generally, an AHU introduces a specific amount of outdoor air (OA%) into the return air, which is then termed mixed air. This mixed air then proceeds through some combination of heating/cooling coils and fans and is then supplied to the space.

An economizer is an operation that introduces more outdoor air than is required by ventilation during times when doing so decreases cooling load. Think a nice 50degree day where the air handler could either use more outside air to help cool the building down, or use a chiller to cool down the return air. There are of course some caveats, but I'll address some of those details in a later blog.

Economizers require dampers, actuators and controls. Any of those three items can be installed incorrectly or fail, leading to non-optimzed economizer function. Because this failure occurs upstream of coils, the occupants are typically none the wiser since the coils step in and provide the correctly tempered air, masking a large energy waste. I'd love to be proven wrong, however, so if you recieve a work order / call from an occupant who states their economizer is not functioning, I'll buy the beers while you tell me the tale.

One method of quickly diagnosing air-side economizer systems is to analyze trend data to flag when systems are not producing the correct mixed air temperature. This, in essence, is peering into the system upstream of the coils and fans, and seeing what the occupants can't. It should be noted that the analysis outlined below is best done once temperature sensors have been calibrated. As a first step I verify that the mixed air temperature (MAT) is bounded by the outdoor air temperature (OAT) and return air temperature (RAT). Since the temperature differences between these values can approach typical error tolerances for sensors, its important to do some sanity checks first.

Once you're confident your sensors are calibrated, proceed to logging data, or pulling trend information from a DDC system. A plot of MAT - RAT vs. OAT - RAT, while awkward to say or type out, is the easiest way to investigate the performance of your economizer operation. First of all note that the slope of a trendline in this graph will produce the outdoor air percentage that was alluded to in the introduction. Doing a simplified energy balance will convince you of this:

OAT*(%OA) + RAT*(1-%OA) = MAT

Flip those values around and solve for %OA and you'll verify the statement above. The real trick now is to develop what the ideal economizer function would look like for this system. For this, I'll introduce a few more terms. The economizer cut-in is the point at which the minOA introduced to the RA stream will exactly meet the supply air setpoint. Generally speaking this is arrived at via the following expression:

cut-in = ((SAT_sp - FanHeat) - (1-minOA%)*RAT_avg))/minOA

Where:SAT_sp is the discharge setpointFanHeat is the heat added by the fan (measure it directly, or just throw in something like 2 degF)minOA is the code required minimum outdoor air flow %RAT_avg is an average return temp.

The cut-in value is usually very low. For a recent system I've done it was 17 degF. This means that at 17degF, you could bring in the minOA%, mix it with (1-minOA%)*RAT, and arrive directly at the discharge temperature setpoint. This is somewhat unrealistic since at that OAT we're probably not actually cooling, but it sets the bounds of how to describe the rest of the economizer operation. Below is a method for calculating the ideal MAT for each sceanrio:

If OA-T is below the cutin --> MA-T = minOA%*OAT + (1-minOA)*RAT

If OA-T is below SAT_sp+FanHeat --> MA-T = SAT_sp+FanHeat. MA-T is then constant in this range. This corresponds to the dampers modulating to meet setpoint.

If OA-T is below RAT--> MA-T = OA-T. This is 100% OA-T. With the exception of the very intitiation of this part of the curve, its interesting to note that *anytime* the economizer is providing 100% OA-T, we typically are employing the cooling coil as well. Think this through, it can be unintuitive if you're using to calling the whole economizer cycling "free cooling". It's just.... "cheaper cooling". Note also that operation of the cooling coil here is termed 'integrated economizer'. For some configurations (think chilled water plants that don't do part load well) this mode is difficult, and is therefore "skipped" and minOA is established once more, and an artifically high, but more easily achieveable load, is placed back on the chiller.

If OA-T is above RA-T--> we revert back to MA-T=minOA%*OAT + (1-minOA)*RAT

Note: if investigating a system with supply air reset based on OA-T, that can easily be integrated into the formulas above by making SAT_sp a formula rather than a constant. If the reset is based upon something else it's probably easier to isolate where the setpoint should be during the economizer operation, and concenrate efforts on trend data in that region. You could even filter out your data into different bins based upon supply air setpoint and produce a curve for each bin. The noise seen in the graph below, where the blue dots have some error above and below the idea line is likely due to the fact that the supply air setpoint for this system was reset by room temperature. I couldn't produce a godo regression of room temperature against outdoor air temperature, so I just used the bottom of the reset since this is most applicable for cooling analysis.