For years, the seismic concerns of legislators and building design professionals were limited primarily to structure-related life-safety issues. That changed after an earthquake considered moderate in seismological terms caused death and widespread destruction in San Fernando, Calif., Feb. 9, 1971. Since then, the focus of lawmakers and designers has been on non-structural systems as well.

This article will discuss the role of legislation in the evolution of seismic design for mechanical systems, as well as the situation arising from the increased stringency of model codes and how the local model-code-adoption process can help alleviate it. Additionally, the article will discuss key conceptual issues that may not be addressed by codes and offer examples of solutions.


A year after the San Fernando quake, the California Hospital Seismic Safety Act was introduced, requiring the state to develop seismic-design regulations for mechanical systems. Little known is the fact the legislation used the phrases “in so far as practical” and “reasonably capable” in describing the goal of ensuring hospitals remain “functional” following an earthquake. The codes that resulted established criteria, but stopped short of prescribing solutions, leaving mechanical engineers to develop their own. For example, on hospital projects during the mid-1970s, we developed calculations and seismic-restraint details for every size and type of pipe and every conceivable situation, even a normally empty pipe being full of water. Because each calculation and detail had to be checked, a bottleneck developed in the design, plan-checking, and construction process. To streamline plan-checking and enforcement in the field, industry standards and guidelines were established and, in some cases, pre-approved. In the process, seismic design for mechanical systems moved from homemade solutions to off-the-shelf details and products.

Recently, procedures for determining seismic-hazard level, importance factor, etc. began appearing in standards and guidelines. Mechanical engineers are expected to follow these procedures and include the results in specifications. Neglecting to do so may put engineers in the position of failing to comply with the very standards they specify.

Still, codes, standards, and guidelines can cover only so much, and some owners have criteria more stringent than those established by code. This leaves design teams to set project-specific goals, define the needed functionality of systems after a seismic event, and find appropriate solutions.


When model codes began addressing the seismic design of non-structural systems, they did so in the “in so far as practical” spirit of the California code. Lacking the constraints of legislation, however, they have become more stringent with each revision. With increased stringency has come increased construction costs — increased to the point many essential-building projects may never get off the ground. Take hospitals, for instance. More and more, health care is being provided outside hospitals, in medical office buildings, which are built to lower standards and, thus, may not be functional following a seismic event.

It is important to remember that model codes are applicable only where adopted. But while model codes can be adopted in part and with amendments, they often are adopted in their entirety, with little thought given to local impact.


Issues not covered in standards and guidelines often are not covered in contract documents, resulting in construction disputes. Some engineers specify that the contractor be responsible for engineering and construction issues not covered in standards and guidelines. This makes bidding difficult because scope is not defined fully. As a result, a bidder may include an allowance that is insufficient or excessive. For that reason, owners and design teams should identify an approach to handling situations not covered by standards and guidelines and define that approach in contract documents.


Some jurisdictions consider manufactured equipment, such as an air-handling unit (AHU) with fan, coils, and filters, a “black box.” Seismic restraint is thought to apply to the box only — what is inside is assumed to be fine. The thinking is that, during shipment, equipment experiences acceleration forces greater than those experienced during an earthquake; if equipment survives shipment, then, it is likely to survive an earthquake. Others argue that because fans do not rotate and coils are not full of water during shipment, the shipment “test” is insufficient.

If a fan is installed on resilient mountings inside an AHU, the “black box” is considered open, and seismic restraints are required both inside and outside.

Sometimes, a “black box” is opened unintentionally. For example, on one project, manufactured AHUs were value-engineered into built-up units. The enforcing agency required seismic calculations and details for all of the components, as well as their attachment to the structure. Opening this “black box” was like opening a Pandora's box, as it resulted in a construction dispute.

Recently, model codes opened “black boxes” wide by requiring equipment to be tested on a seismic shaking platform. Although, in many cases, equipment certified as having passed such testing is not yet available, local jurisdictions have adopted the requirement. This illustrates the need for selectivity in the model-code-adoption process.

Although the “black-box” concept is acceptable for many types of buildings in California, it has been rendered useless in many less seismically active parts of the country thanks to the blind adoption of model codes requiring the shaking test. This is unfortunate, as there is no reason why seismic codes should be more restrictive, say, on the East Coast than they are in California. Through active participation in the public-review process, engineers can help ensure the appropriateness of model codes for their jurisdiction.


The seismic bracing of mechanical components imposes force on structures at the point of attachment. The capacity of structures to withstand this force is the responsibility of the structural engineer. Mechanical engineers should work closely with structural engineers to establish proper seismic-design and attachment criteria. This entails:

  • Determining whether seismic bracing for suspended pipes can be attached to the floor above or to structural beams only. In either case, maximum forces should be established.

  • Coordinating the force applied, the type of concrete, and the characteristics of the concrete, which determine the size and embedment depth of anchor bolts and expansion anchors.

  • Establishing procedures for testing anchors in the field and coordinating and monitoring that testing.

Extra coordination is required when working with existing structures.

Parameters for seismic bracing and attachment to structures should be established during design, with specific coordination coinciding with the development of construction shop drawings, when all details are known. Most seismic design cannot be coordinated fully during design. Reasons include specification options and contractors' preferences, such as those concerning:

  • Pipe material (steel or copper).

  • The actual weight, center of gravity, and support configuration of equipment.

  • Duct type (rectangular or round).

  • The location of pipe and duct supports.

  • The routing of electrical conduit shown as home runs.

  • The size and location of sprinkler-system piping.


In pipe systems, seismic bracing is applied to pipe supports at regular intervals — say, every 40 ft — with unbraced supports in between. During the 1970s, seismic bracing was performed almost as an afterthought. The same supporting rod and nuts used for pipe support were used for pipe bracing. In many cases, pipe could not be braced diagonally until it was loaded fully and the slope adjusted and confirmed. The pipe support used for seismic bracing then had to be taken apart to incorporate the diagonal bracing. In time, details and products allowing pipes to be braced without supports being taken apart were introduced.

In recent model codes, seismic-restraint requirements for mechanical systems differ by building type, type of support (resilient or rigid), even elevation. For example, the seismic-restraint requirements and, thus, details of a 4-in. pipe suspended at the penthouse level of a building are different than those of a 4-in. pipe suspended in the basement. Seismic design also may differ by importance factor. For example, two 4-in. pipes suspended next to each other may have different seismic-restraint requirements and, thus, details because of different importance factors applied to their systems.

Seismic-restraint guidelines reference many details and tables. Because this data may be difficult for installers to use properly in the field, some seismic restraints are planned in greater detail during the shop-drawing phase. Bracing points of pipes and ducts are identified, seismic-restraint components for each bracing point are selected, and the components for each point are shipped separately. Such coordination reduces guesswork in the field.


Many buildings include seismic separation joints. Because a building will act differently on one side of a separation joint than it will on the other during a seismic event, engineers should minimize the crossing of separation joints by pipes and ducts. When such crossing is unavoidable, flexible pipes and ducts should be used, and the magnitude of movement of the two sides of the joint should be defined in contract documents. The magnitude of movement is a function of the elevation of a crossing point and should be available from the structural engineer.


When base isolation is used under a structure, acceleration forces inside the structure are reduced. However, codes may not allow one to take credit for the reduced acceleration forces on the mechanical system. On the other hand, movement of building sections may be greater in a base-isolated structure than in a conventional one, making the crossing of building seismic separation joints more challenging.

Air space between building sections compresses or expands as building sections move, producing air pressure that can damage building components. Pressure-activated swing panels are one means of alleviating this pressure.


In seismic restraint, equipment is connected to a structure rigidly. In vibration isolation, it is installed on devices designed to prevent the transfer of vibration to a structure. This conflict of objectives can be resolved through close coordination among acoustical, structural, and mechanical engineers.

For example, on many projects, the same vibration-isolation details, including inertia base, are used for all pumps, whether they are installed on an upper floor with occupied space below or on slab on grade in a remote central plant. Inertia-base weight is a major factor in seismic-restraint design. Most likely, vibration isolation for pumps installed on slab on grade in a remote central plant can be relaxed, reducing the initial cost of both vibration isolation and seismic restraint. Vibration isolation should be application-specific in its design to reduce complexity and cost.

Often, specifications require resilient support of pipes and ducts. Seismic-bracing criteria and details tend to be more elaborate when pipes are supported resiliently. Because of complexity and cost, where possible, resilient support should be limited to short distances.

Charles Salter, PE, a leading acoustical consultant, advises engineers to ensure that designs meet Occupational Safety and Health Administration hearing-damage-risk requirements. Also, he encourages them to analyze the risk of repairing vibration isolation if an installation fails to meet project criteria.


Some buildings or portions of buildings have dual purposes, including to provide refuge after an earthquake. That secondary purpose — and the higher importance factor that may accompany it — is not always apparent, however. For instance, on one project, a conference room's intended use as a command center in emergency situations was not discovered until late in the design phase. Because of a heightened importance factor, the HVAC system had to be redesigned. Thus, early disclosure of a building's intended use in an emergency situation is key to success.


Often in campus settings, essential and nonessential buildings are served by the same mechanical system. With essential buildings expected to perform better than nonessential ones during seismic events, a failure in a nonessential building could render a complete mechanical system inoperative. Therefore, system design goals should be defined, situations analyzed, and solutions developed. For example, to maintain the functionality of a campus hot-water system, automatic isolation valves controlled by pressure sensors in each building system could be designed. A drastic drop in hot-water pressure in a building would close the automatic isolation valves serving that building, preventing drainage of the entire campus hot-water system. This approach would have the added benefit of protecting essential buildings from pipe failures unrelated to seismic events.


Pipes serving seismically compliant essential buildings should be routed through structures designed to similar seismic criteria. Otherwise, the failure of a seismically non-compliant structure could render the system nonfunctional at an essential building.

A few days before patients were to be transferred to a new, seismically compliant wing of a hospital, several pipes feeding the new wing were discovered to have been installed through a seismically non-compliant existing wing. Finger-pointing ensued after inspectors prohibited the transfer of patients to the new wing. After much debate and a long appeal procedure, the inspectors were convinced that patients would be safer in the new wing, where the greatest danger was of water services being cut off in the event of an earthquake, than in the existing one, where they risked something falling on them. A compromise was reached: The patients were allowed to move to the new wing in exchange for the owner agreeing to install new pipes within one year and provide quarterly progress reports. Although this was not an inexpensive solution, it kept the safety of patients in mind “in so far as practical.”


Seismic-design criteria are becoming more stringent in model codes, and the “in so far as practical” spirit is fading. In addition, many local jurisdictions do not recognize how stringent the model codes have become and are not being discerning in adopting them. On the other hand, codes and standards cannot cover all situations related to the seismic design of mechanical systems. This challenges design teams to define project-specific goals and find appropriate solutions.

For HPAC Engineering feature articles dating back to January 1992, visit

The recipient of several local and national engineering-excellence awards, Shlomo Rosenfeld, PE, is president of Shlomo I. Rosenfeld & Associates. From 1987 to 1997, he was the mechanical-engineer member of the California Hospital Building Safety Board. He can be contacted at