Talking with mechanical engineers, I find a range of criteria being used to determine the cooling load of electrical equipment spaces. Estimates of cooling loads based on equipment efficiency range from 1.5% to 5% of the kW being served by the space. Others use rules of thumb for cfm/sq ft, or W/sq ft. All would prefer to receive calculated values from the electrical designer, but they usually don't.

In many cases, electrical equipment load is insignificant with respect to other space loads, and proven rules of thumb are reasonable and cost effective. However, in large, concentrated installations such as substation rooms and data center back-of-house, loads are significant and the cost of errors can be high, so I would make a case for more accurate load calculations.

Sources Of Information

The heat generated by electrical equipment results primarily from inefficiency or losses. Manufacturers can provide values for heat loss in switchgear, bus, motor starters, switches, and circuit breakers; these values typically are based on each component's continuous current rating.

Losses for power conversion equipment such as vfd's and uninterruptible power systems can be calculated from published efficiencies. In the case of process equipment or utilization equipment also located in the space, there is often no clear correlation between nameplate ratings and losses and actual heat generated; obtain this data from the manufacturer.

Consider The Load

The majority of losses in electrical equipment results from power dissipated by current (I) flow through the resistance (R) of the bus bars, wires, and other conductive components. Heat produced is proportional to the square of the current (P = I2R), and is commonly referred to as "I-squared-R" loss. Thus, as the load is reduced, the heat generated is greatly reduced; a bus bar carrying 75% of rated current will produce only 56% of the heat it would produce at rated load. Values reported by manufacturers for heat generated in bus, circuit breakers, and other distribution equipment components are typically reported for the full rated current of the component, and can thus be adjusted to reflect the actual load.

Most equipment does not carry full rated-load current even at peak demand periods. Most circuit breakers and fusible switches are only approved for continuous operation at 80% of their rated current when installed in an enclosure (as almost all commercial electrical equipment is). Further, the method by which loads are calculated under the NEC is conservative, which results in significant initial oversizing of service equipment for a typical commercial building. Adding up the full load losses for all of the components in a space may provide unreasonably conservative values that lead to oversizing of systems.

When I have a high concentration of distribution equipment in a small space, I typically use a spreadsheet-based calculation that modifies the full load loss values for each component to reflect I2R effects at actual load.

Transformers Are A Special Case

Transformer loss is divided into core loss and load loss. Core loss is magnetizing loss, independent of load, and exists whenever the transformer is energized. Load loss is I2R loss and varies with load as discussed above. Both values can be obtained from the manufacturer and calculating heat generated by a transformer requires both.

Accurate estimates are important because transformer losses often dominate the cooling load in an electrical room with major impact on system sizing and ability to meet design conditions; and because they may vary with the design of the transformer, rules of thumb are unreliable. Indoor dry-type transformers are available in three different designs, based on different full-load winding temperatures. Full-load efficiencies between these designs can vary by a full two percentage points for transformers of the same size and manufacturer, and they can vary by as much as four percentage points across the full-size range for a single manufacturer.

Don't Forget Contingencies

Finally, don't let accounting for actual load lead you to overlook contingencies that may represent the worst-case scenario. A good example is the common "double-ended" substation with a secondary tie switch. It is common to load the transformers so the total load is within the 133% force-cooled capacity of a single transformer, providing the ability to take a unit out of service without affecting the load. Normally, load is divided evenly between two transformers (each loaded at 67%), and the total load loss would be 89% of the rated load loss of one transformer. When the entire load is switched to a single transformer, now loaded at 133%, the total load loss would double to 177% of the rated load loss.ES