Emergency power for today’s healthcare
There is an even greater need for hospital services during the types of emergencies that cause power events. And it's a major challenge to the design team to determine the requirements for the emergency power system. As in all engineering projects, it helps to break down the design into individual tasks and decisions.
There is an even greater need for hospital services during the types of emergencies that cause power events. And it’s a major challenge to the design team to determine the requirements for the emergency power system.
As in all engineering projects, it helps to break down the design into individual tasks and decisions. The tasks consist of identifying potential power problems for the hospital, determining the essential system requirements and loads to be served, designing the emergency system, and protecting the system. The capital available for these projects is limited, so for each step, the system design must be evaluated on a cost basis to identify needs versus wants.
It’s important to get in touch with the utility provider early in the project to determine some of these items. The utility staff can help identify the answers to some of these questions.
For example, one hospital client, located in a high-growth area, is on a grid that is expected to have the addition of two commercial supercenters, one office quad, and a high-rise residential building, all within a year. At the same time, the hospital was poised to double in size. Working with the utility, engineers were able to determine that all of these buildings would be fed from one substation. However, the utility was pulling all new dual feeds to each, and replacing the wiring to the hospital with a loop feed. Also, the hospital was given the option to expand the system, for a fee, to be fed from two substations.
Some of these decisions are based on the probability that a failure will occur in the normal system. Unfortunately, natural disasters also play a part. Tornados, hurricanes, floods, and lightning can have an impact on the operation of the normal service.
Hospitals also affect the reliability of the distribution system. Neglect reduces the lifespan of equipment and increases the likelihood of problems, whereas a preventive maintenance program, with regular testing of equipment and infrared testing of service gear, will help catch some of equipment issues before an operational failure occurs.
What is on emergency power?
The second task requires collective effort of the owner, operators, users, and sometimes the local authority having jurisdiction (AHJ).
NFPA 517, 99, 101, and 110, the state and local codes, and the AIA Guidelines for Healthcare Facilities all provide some input on what must be connected to emergency power. This is the easy part. The question is, what other items does the owner want on emergency power? Does the facility want the complete HVAC system on emergency or just some areas? What about radiology equipment? These are not required by code, but the owner may want to use them in an extended outage. To what extent will the kitchen operate? Does the facility intend to maintain normal operation or has the owner established a plan to maintain food services during a power outage? Are administrative spaces required to be on emergency power?
There is a growing trend in the healthcare industry toward backing up the entire hospital. Often this is an owner-driven request, because there are benefits the owner receives from backing up both the normal and essential services. Many utilities have programs where hospital owners receive financial payback on their generator systems. Several of these programs are designed to get the hospital off the grid when the utility needs the additional capacity, and the hospital gets paid based on the capacity of power they remove from the grid. Hospitals may even receive some of the capital costs for the energy plant upgrades from the utility company.
Providing an emergency generation system that provides power to the entire facility in an outage provides additional benefits for the entire system. The design still maintains a normal branch of power; in this case it could be defined as the nonessential branch. This branch includes loads that typically would be connected to the normal branch and have their own transfer switch. These loads do not require emergency power per code. Therefore, power to these branches will be restored last. In Figure 1, these loads are represented as priority 400 loads, automatic transfer switch (ATS) N1 through N4. Also in the event that one generator is out of service, or there is insufficient power, the non-essential branch can be shed. Transfer switch schemes along with distribution design enables the control of load priorities, including these normal loads. For example, Figure 1 shows a system where the radiology equipment is a priority 2, just a step behind the code-required priority 1 branches.
Figure 1 is an example of a normal/emergency transfer switch priority scheme using multiple transfer switches. Only four transfer switches are shown in the one line for clarity, but the figure includes all transfer switches within the system. The purpose is to break down the loads to determine what load is connected to the bus first. Priority 1 is the branch that must have power restored within 10 s, the life safety and critical branch loads as well as generator equipment loads required to be on in 10 s. The priority 2 loads will come on second, after a pre-set time delay as long as the second generator is running. This will repeat again for the priority 3 loads and the third generator, then again with the priority 4 loads.
At this point in the design, the emergency system is starting to take shape. We have clearly defined how long we may be without power, and what additional equipment, beyond code-required equipment, the owner would like to include in the emergency system. It is time to start the emergency system design.
Design of the emergency system
The third stage of the essential system design is to determine the emergency power source, elements required for the essential system, and fuel storage requirements. State and local AHJs have requirements. For instance, Florida requires a diesel generator with enough fuel for 64 h of operation with the generator loaded 100%. However, NFPA 99 only requires sufficient fuel to operate the emergency system for 90 min.
This also brings us back to the first question: How long can the hospital operate without normal power? An anticipated outage time of a few hours would require a limited amount of on-site generation. If the anticipated outage is a day or more, the size and scope of the emergency power system must provide emergency power for even more of the facility. Some designs use UPS systems, but this solution is limited because it requires the normal service to recharge the batteries. A fuel truck can deliver more fuel just before fuel runs out, but UPS systems are limited. A UPS system should be used for equipment that cannot tolerate interruption of service. The UPS system should be backed by the emergency generator system.
Typically, diesel generators are used for emergency systems. While you determine your loads in step 2, determine which loads are priority 1, non-shedable loads. These loads are required to have power restored in 10 s. What is the largest generator that can accomplish this? Engineers must consult with the engine manufacturers to determine the appropriate solution. The fuel source will affect the size; for instance, a 1,500-kW diesel generator will be able to restore power in 10 s, while a 1,500-kW propane generator may not.
Sometimes the project itself will determine the number of generators needed. For example, a hospital in Florida has a parallel system with three 600-kW gensets, and space for a fourth 600-kW genset. The existing generators are operating at approximately 95% capacity, but the new loads to be added as part of an expansion are estimated at 1,000 kW. The new load is larger than the potential emergency system expansion, reducing the available options for the expansion. The design can either include new parallel switchgear rated for five 600-kW generators and two new 600-kW generators or a stand-alone generator unit rated for 1,000 kW. The cost-effective solution will be to provide a new stand-alone generator. The existing parallel switchgear has limited flexibility because the size of the future generator and future expansions are defined when the parallel switchgear is installed. It also is limited by the bus rating, or the space in the room that an expansion section would take up, or maybe the fault rating. In this example, the original design allowed the addition of a 600-kW generator, but no greater. In order to increase this capacity, the parallel switchgear would have to be replaced.
Usually it is not as straightforward as this example, so engineers must decide which direction to use based on their judgment. More often than not, the essential load will require the use of multiple generators in a parallel system.
The parallel generator system is more reliable than a single generator system, allowing the use of multiple generators to provide emergency power. There are a few limitations to parallel generator systems, such as priority 1 loads and increased fault current levels. Priority 1 loads have to be on in 10 s, and therefore are limited in size. When an outage occurs, a signal is sent to all generators to start. The first generator to come up to full speed will be connected to the bus, then the remaining generators will sync to the first generator and connect to the bus. For priority 1 loads the first generator must be on in 10 s. The first generator on could be any of the generators, therefore the priority 1 loads are limited by the size of the smallest generator. Priority 1 loads will include life safety branch, critical branch, and some equipment branch loads such as generator equipment.
One can use a tie breaker in parallel switchgear designs. As shown in Figure 2, the tie breaker in the open position will allow for the “A” side generator to come up to speed and connect to the bus at the same time as the a “B” side generator. This effectively doubles the priority 1 load capabilities of the system because each side can be on in 10 s. Once they are on, the generators can synchronize and the tie breaker closes, creating a single bus, or the tie can remain open if design requires it.
Typically the emergency distribution system is not the primary contributor of fault current, but when multiple generators are paralleled together, the fault can exceed that of the utility service. It is important that design engineers evaluate the fault both from a normal service and an emergency service standpoint. The example shown in Figure 2 is taken from a project where the emergency side generator fault (including the estimated fault contribution from the two future generators) is approximately 90,000 amps, significantly more than what is available at the utility service entrance.
To further clarify the performance of a generator in a fault scenario the engineer must become familiar with the generator decrement curve. The decrement curve represents the fault output of a generator and is plotted as a function of various generator characteristics including subtransient reactance—Xd” (one to five cycles), transient reactance—Xd’ (five to 200 cycles), synchronous reactance Xd (above 200 cycles), subtransient time constant (Td”), transient time constant (Td’), armature short circuit time constant (Ta). The reactances are typically per unit values with the generator kVA and full load amps as the base. These values should be provided by the generator manufacturer, along with the a plot of the generator decrement curve. This document also should include the full load amperage of the generator, and the instantaneous fault output. Now that we have these values it is easy to determine what the instantaneous fault of the generator is. The equation for the instantaneous generator fault is Isc =Ea / Xd”, where Ea is the internal generated voltage identified a 1.0/ unit, simplifying the formula to Isc = 1 / Xd”. We should we evaluate instantaneous available fault current within our calculations because the automatic transfer switches and electronic trip breakers we use typically only have a 3 cycle withstand rating.
After the initial fault output the generator fault output will decay rapidly. In order to maintain a fault current substantial enough to isolate the faulted breakers, permanent magnet generators are used. A permanent magnet generator will maintain 300% of the full load output of the generator for approximately 10 s. Engineers often use the full load current times three to determine the fault output of the generator. However, the fault current created by the generator will still be greater within the subtransient period. It’s important that the designer consider this when paralleling multiple generators together.
Protection of the emergency system
NEC 2005 states that legally required standby systems’ overcurrent devices should be selectively coordinated with all supply side overcurrent protective devices. The emergency distribution system for healthcare must be designed in such a way that the system is selectively coordinated.
The goal of selective coordination is to make sure a fault or overload in one location of the system does not cause power loss at other locations in the system. An elementary example of coordination is the following. Imagine a branch circuit panel board, with a 100-amp main breaker, and a 20-amp branch circuit breaker serving lighting an operating room. A problem on the lighting circuit should trip the 20-amp branch circuit breaker before it trips the 100-amp main breaker and causes the loss of power to the remaining circuits. It is the responsibility of the designer to make sure the breakers used for the 100-amp breaker selectively coordinate with the 20-amp breaker. Normally the 100-amp and 20-amp breakers are molded-case, non-adjustable breakers. Larger breakers come with electronic trip units, which are adjustable. Here the engineer must carefully adjust the settings to achieve satisfactory coordination.
The NEC does not clearly define what must be selectively coordinate. Therefore, it is up to the engineer to work with the local AHJ to determine an interpretation of what is selective coordination and at what the requirements will be. In Florida healthcare, the local AHJ has required selective coordination plots, signed and sealed by the engineer of record, for years. The requirements for selective coordination have evolved through the years in this region and currently require selective coordination from 0.10 s on.
In order to understand the protection of the generator and the selective coordination of the generator, some basic understanding of time-current curves is required. Time-current curves are log graphs that represent the tripping time versus the current of breakers. They also are used to demonstrate the damage curve for generators, transformers, motors and wires, and to evaluate the fault contribution from a generator, or the inrush seen from a transformer. To do a complete selective coordination study, all of the above items must be considered.
For a generator the two important components are the damage curve and the decrement curve. The damage curve (i.e. overload curve or withstand curve) is a measure of current from which the generator will be damaged. The decrement curve, as stated above, is a measure of the fault current produced by the generator. In the event that the generator faults, we want the breaker closest to the fault to trip and protect the system. The fault current must be sufficient to allow the breaker to trip. A standard generator cannot maintain fault current for long, so standard practice is to use a permanent magnet generator. This generator will maintain approximately 300% of the rated current for approximately 10 s, long enough for the downstream breakers to trip and protect the system. At approximately 10 s, the decrement (fault) curve will intersect with the damage curve for the generator. The selective coordination for the emergency power system must protect the generator, but allow the downstream breakers to trip.
Figure 3 shows a panel serving two ATS. Panel ESB is the parallel switchboard (only one generator is shown) that serves a transfer switch ATS-EQ via a 1,200-amp breaker. This board is fed by several 1,500-kW generators with 2,500-amp breakers protecting each. Selective coordination requires that the ATS-EQ breaker trip before the GEN breaker.
But if the fault is greater than both breakers, what will happen? We want to avoid a situation where a fault will take a generator offline before it takes the faulted branch of line. In Figure 3, we don’t want a fault in ATS-EQ to take the generator offline because it will remove the power for ATS-CR. This is not selective coordination.
The goal is simple: design the protection that allows the downstream breakers to trip before the generator breaker trips. The generator breaker in Figure 4 is set to the right of the decrement curve from 0 s through 2 s, and crosses to the left of the decrement curve before damage to the engine occurs. Remember, the decrement curve is the fault output for the generator. It is important to note that many generator controllers have internal overcurrent protection that will take the generator off-line as well. Make sure you work with the manufacturer to confirm you have a complete selective coordination system.
Emergency system design is a critical first step toward the proper operation of today’s healthcare facilities. Identifying the issues and addressing the challenges one at a time, we will be able to provide a safe and reliable emergency system, and do our part toward enhancing patient safety.
|Cost||Single point of failure|
|Less space required||Future growth is limited|
|Increased generation capacity||Cost|
|Future growth||Fault levels|
|Control schemes||Limited flexibility|
|Tie breaker schemes|
What are the risks?
The first question is: What is the likelihood that at some point in the life of an emergency power system, a hospital will need to operate on generator power for several hours, days, or even weeks? Normal service must be evaluated for reliability and the following questions are helpful in determining the likelihood of a normal power outage:
• What is the history of power problems in the area? Typically the local utility can provide a history of what outages have occurred and the reliability of the services.
• Does the hospital have dual services, and if so, are they from separate substations?
• What else is on the grid with the hospital and how far is the hospital from the substation?
• Is the hospital located in a high-risk area for earthquakes, tornados, or hurricanes?
• Is the area prone to lightning strikes?
• How well does this particular hospital maintain its equipment?