How do I calculate the backup power requirements of my facility? Answering this question requires not only knowing what loads to backup and what their power requirements are, but also understanding how different types of loads affect the backup power source, how to manage the connection of loads, and the constraints building codes may place on this process. Although this article addresses some aspects of different types of backup power supplies, it will focus on the most common backup power supply, a standby generator set in which a liquid or gaseous fueled reciprocating engine is coupled to an AC generator (alternator).


Come to Terms with the Load

Let’s start with basic terminology used to describe electrical loads and how it relates to the capabilities of backup power sources. Loads can be defined in terms of real power and apparent power. Real power, measured in kilowatts (kW), is electrical energy provided by the source and consumed in the load; it can be thought of as power that produces useful work like pumping fluid against pressure or raising an elevator against the force of gravity. Apparent power, measured in kilo-volt-amperes (kVA), is the arithmetic product of voltage and current and represents the total current capacity that the source must have. The difference between these, reactive power (kVAR), creates the magnetic fields necessary for machines like motors and transformers to operate but is not converted to work or dissipated in the system. Power factor (PF), defined as the ratio of kW to kVA, is then a measure of how “efficiently” the system uses current-carrying capacity to provide useful power to the load.

This distinction is important to the standby power source. In a generator set, the engine produces real power, and must provide mechanical power output at least equal to the maximum load kW plus the losses in the alternator. The alternator converts mechanical power to electrical kW, but also supplies the reactive power component and must therefore have current-carrying capacity at least equal to the maximum kVA. In battery-based systems the battery stores energy and must be sized for the kW requirement and discharge duration, but the power electronics equipment (inverter) is rated on current-carrying capacity and must be able to provide the required kVA.

Another important distinction is between average and peak load. Average load determines the required energy storage capacity of a backup power source, such as fuel consumption for a generator set or discharge time of a battery bank. Peak load, on the other hand, determines equipment ratings: an engine must provide mechanical power output to supply the peak kW even if the average kW is only half that value. Likewise, an inverter must carry the peak kVA without overheating or shutting down regardless of the average value of the load. 

If planning to back up the entire load of an existing facility, you may be able to obtain the average load data from your electric utility. Utility bills report energy consumption in kilowatt hours (kWH) and average kW can be determined by dividing kWH usage over time by the time period. Beware, however, of using kW demand reported on utility bills as equivalent to peak kW for sizing the backup power source. Many utilities calculate kW demand by summing kWH over a set period, the demand interval, and dividing by that period. Thus reported peak demand is actually the highest average kW recorded over any demand interval, which may be as long as 15 minutes. If loads in your facility cycle on for periods of less than twice the demand interval, peak kW seen by your backup power source may exceed the utility’s reported peak demand.


All Loads are Not Created Equal

The next step is to categorize your loads based on their electrical characteristics. In a large system, such as a campus central power generation plant, load diversity may make this step unnecessary, but in most facilities it must be considered because different types of loads have different impacts on sizing of the backup power source. Two load categories that have significant impact on power supply sizing are large motors (or groups of small motors that start simultaneously) and “non-linear” loads such as uninterruptible power supplies (UPS) and VFDs.

Large motors started at full voltage draw six to eight times their full load ampere (FLA) rating while accelerating. Current stays at this high level, referred to as locked rotor current, until the motor reaches 80-90 %of rated speed, when it begins to drop off and ultimately settles at or below the FLA depending on the power required by the driven equipment (Figure 1). Locked rotor current is mostly reactive and does not affect the real power the system must supply, but has implications for the standby power source: First, the equipment must be able to carry this current for the acceleration period without overheating or tripping, which may be a concern for power electronic equipment such as inverters used with fuel cells and battery-based energy storage systems; second, anytime you draw current from a power source the voltage dips due to voltage drop across the source impedance, and since standby sources have higher impedance (are less stiff) than utilities, voltage may dip during starting to a level that can’t be tolerated by other loads on the system.

Figure 2 shows typical generator voltage response to large motor starting. Voltage dips instantaneously to a minimum determined by starting current and alternator characteristics followed by recovery to a higher level as the voltage regulator responds to the transient. A corresponding frequency dip will also occur as the engine and governor respond to additional load. Once the motor has accelerated both voltage and frequency return to steady-state. Acceptable instantaneous voltage dip depends on the tolerance of the loads served; 20% is a commonly used limit and 35% is generally considered the absolute maximum to prevent on-line motors from dropping out. Frequency dip is usually less of a concern. Manufacturers have some ability to improve the motor starting performance of a generator set by providing an over-sized alternator or varying voltage regulator characteristics.

Even supplied from the utility, voltage dip on starting a large motor may be excessive and many methods of reducing starting current, generally by reducing motor voltage during starting, have been developed (Table 1). You should determine whether reduced-current starting methods are used for the motors in your facility when incorporating them into the calculation process. However, before you consider changing the starting method of an existing motor to reduce starting current be aware that reduced starting torque may result in insufficient margin between motor and load torque to accelerate to rated speed; consult the motor and driven equipment manufacturers before converting an existing motor to a different starting method.

We use the term “non-linear” to describe loads which draw current under the control of internal switching devices. When loads that consist only of resistance, inductance, and capacitance such as motors and heaters are supplied directly by a sine wave voltage, the current is also a sine wave with the same frequency as the voltage (Figure 3). However, current drawn by loads using power electronic switching such as UPS systems, computer equipment, and VFDs is not a simple sine wave at line frequency, but a waveform that is the sum of a series of sine waves with frequencies that are multiples of the line frequency called harmonics.

Non-linear loads have two effects on backup power sources: First, harmonics create additional heating in current-carrying components like generator windings; second, harmonic currents can distort the output voltage waveform, which may adversely affect other loads on the system, as shown in Figure 4. This distortion is quantified as Total Harmonic Distortion (THD), expressed as a percentage of the fundamental frequency voltage. For systems serving a mix of load types, a limit of 5% THD is recommended; systems serving only non-linear loads may be able to accommodate levels up to 10%.

Another category to consider is loads that may act like a generator and attempt to return power back to the source. This “regeneration” is sometimes used as a means of braking in hoists and elevators and can wreak havoc with a system that is not designed to absorb the power, causing voltage or frequency transients and circuit breaker tripping. If such loads are used on a standby power system, care should be taken to assure that there is always sufficient other load on the system to absorb the regenerated power without trying to force it back into the source.


Managing the Load

It is advantageous to apply loads to the power source in discrete blocks or steps, rather than connecting the entire load in one fell swoop. Although many backup power sources can accept rated load in one step without failure or shutdown, voltage and frequency transients associated with such block loading may be severe and delay recovery to acceptable voltage. With motor starting, applying load in multiple steps provides less voltage dip per step and better power quality for loads already connected to the system.

Load may be segregated by using multiple transfer switches with different time delay settings, by automatic control from a building automation system, or by manual control of restart after the system switches to the backup source. In buildings where most of the large motors are controlled by VFDs, the inherent acceleration ramp feature of the VFDs has the effect of spreading out the addition of load to the backup source without intentional segregation.

Transient performance aside, large systems with multiple generators in parallel on a common bus that serve loads requiring immediate restoration of power must have their load segregated such that the amount of load that is allowed to connect immediately is within the capacity of the first generator on the bus; additional blocks of load can be connected as subsequent generators synchronize to the first and are connected to the bus.

If the facility is required by code to have backup power, the National Electrical Code (NEC) places constraints on delayed connection of loads. In hospitals, loads that are necessary for life safety or critical patient care functions are required to connect automatically within 10 seconds. Other loads necessary for hospital operation may use delayed automatic or manual connection. In other occupancies, loads designated by the NEC as emergency or legally-required standby require automatic connection within specified times while optional standby loads may use delayed automatic or manual connection.  


Using Manufacturer’s Sizing Software

Web-based sizing software for generator sets can simplify the process of translating a list of loads into an appropriately rated unit. These programs allow you to input site conditions and performance criteria such as limits on voltage dip, frequency dip and harmonic distortion, then create load steps and assign loads in different categories to each step. Load categories available include lighting, air conditioning, motor, battery charger, UPS, medical equipment, and welders. The software then selects one or more generator sets rated to carry the total load under site conditions without exceeding any of the specified limits as each load step is applied.

I used three manufacturers’ sizing programs for an example facility with connected loads consisting of 20 kVA of lighting and miscellaneous power, a 300kW/375kVA UPS loaded to 80%, and three 30-ton DX cooling units. The loads were segregated into five steps: lighting and miscellaneous power, UPS, and the three cooling units individually. The load parameters calculated by the software based on its internal model of each category of load is shown in Table 2, and the resulting generator selection and predicted performance in Table 3.

The three programs agree within a reasonable margin on the parameters of the load, and the small differences there are most likely the result of differing assumptions for specific categories of load, such as cooling equipment EER, motor locked rotor code, etc. Some of these parameters can be modified by the user to reflect actual equipment, but there is minimal return for the extra effort required to be precise in this process, as the effect on sizing will be small relative to applying a reasonable safety margin to the result to allow for future expansion, equipment replacement, and performance degradation over time. Although selected generator ratings are close, predicted performance varies significantly, likely reflecting different alternator, governor, or voltage regulator parameters.

Because of variability illustrated by this example, I recommend avoiding exact generator ratings in a specification or request for quotation. Rather, state a minimum kW rating, identify performance criteria, and provide enough load data to allow the manufacturer to select an engine-alternator-voltage regulator combination to meet those criteria. For this example, with no future growth anticipated, you might specify a minimum rating of 600 kW at site conditions, list load steps and categories, and provide limits for voltage dip, frequency dip, and harmonic distortion.


Trust but Verify

 The final step in the process is to verify both steady state and transient performance of the installed system. While this may be part of the manufacturer’s factory test prior to shipment, it should also be included in the commissioning process as site conditions may affect results.