Networking VFDs in high-performance buildings

Historically, the use of variable frequency drives (VFDs) has been limited to large motors and projects with healthy infrastructure budgets. However, as each iteration of the energy code mandates variable flow control of mechanical systems on smaller systems, VFDs have become more commonplace. While their implementation has been typically limited to providing code-required variable flow, VFDs are capable of communicating a wealth of information on system energy consumption, equipment health, and diagnostics.
By John Gross, PE, ATD, Page, Houston March 15, 2017

This article is peer-reviewed.Learning objectives

  • Realize that variable frequency drives (VFDs) are standard for improving energy efficiency in buildings. 
  • Understand the ability to integrate VFDs with the building’s automation or management system.
  • Learn how VFDs can enhance high-performance buildings.

The past 20 years the commercial building industry has seen tremendous change in countless aspects of building design, construction, and operation. Variable frequency drives (VFDs), once a luxury reserved only for high-budget projects, have become the accepted means for complying with energy requirements for motor-speed reduction on all but fractional-horsepower motors. Traditionally, VFDs are controlled and monitored via simple hard-wired signals, while the extensive data and control available via network integration is left unused. There are good reasons why this has historically been done, but are those reasons still valid? What control and monitoring is left on the table? What value is obtained when network integration is completed? What challenges exist when considering integration?

Historical operational issues

The dynamics of commercial building construction and operations have changed over the past several years. The International Energy Conservation Code (IECC), building management systems (BMS), and building operations have undergone tremendous changes. While the IECC and ASHRAE 90.1—Energy Standard for Buildings Except Low-Rise Residential Buildings have driven the basic energy code requirements, sustainable building design and various green building organizations have pushed the envelope toward sustainable design. 

ASHRAE produced Standard 189.1 for Design of High-Performance Green Buildings, and in 2015, the International Code Council released its first International Green Construction Code. Fundamentally, depending on the climate zone in which a building is located, these codes have driven a reduction in the energy-use index of 39.3% from 1989 to 2010, according to a 2014 review by the U.S. Department of Energy. Through the efforts of countless organizations, sustainability has become part of our way of life—and with it, social pressure on businesses to not only be sustainable, but also to make it part of the public relations message. Though many organizations are choosing not to pursue the formal green building certifications available, many see the benefits that high-performance buildings offer and design new projects with this in mind.

Figure 1: Networking variable frequency drives (VFDs) gives the operator visibility to myriad additional data points for troubleshooting and energy optimization. The additional data, when properly trended, also provide valuable information to technicians in the event of equipment failure. Courtesy: Page

Around 20 years ago, most businesses maintained a staff of building operators—HVAC mechanics, electricians, plumbers, and building engineers—whose sole responsibility was to ensure quality, reliable service to the building occupants. These individuals knew their buildings and the tricks to keep them running. They had tenure at these facilities and knew which individual was always hot or cold. However, recently, more companies have begun to view building operations less as a critical part of the business, and more as unprofitable overhead. In many cases, building operations are now even considered a cost center by some businesses, with pressure to operate as lean as possible, which includes minimizing utility costs. They are frequently viewed as a necessary evil, because without an effective operations team, the occupants cannot perform their jobs. However, the resultant attitude is that many companies have either downsized the operations staff or outsourced their building to the lowest bidder. This results in significant turnover in staff, which also can include a reduction in physical bodies. In either case, facility knowledge is lost, which also includes the ability to diagnose and correct operational issues.

Technical issues

Two decades ago, a major spec issue for BMS projects was whether the system was Windows-based or DOS-based. Capturing meaningful trended data was difficult, as many systems connected the direct digital controls (DDC) controller to the server via twisted-pair serial wiring. Data throughput was a major challenge. Systems now typically operate on a thin-client (Web browser interface), connected to servers with large hard drives and communicate with the DDC controllers via Ethernet cabling. Data throughput and storage is no longer a problem in many facilities. Though not within the scope of this article, it is important to note the extensive use of Internet protocol (IP) communications in a present-day BMS does present a security vulnerability when the BMS resides on the same network as the corporate information technology (IT).

As the commercial energy code, public relations, and a new generation of employees drive for more energy efficiency, tighter construction budgets mean fewer safety factors for infrastructure design. The smaller safety factors mean that less significant operational issues, which in previous years were transparent to the occupants, now create disruptions to the building operations. The view of building operations as a cost center now also means that losses in energy efficiency and occupant disruptions are more highly visible and critiqued.

In short, businesses expect greater energy efficiency with less disruption to the occupants, but with smaller budgets and less staff who have less experience at the facility. So, how does VFD networking help achieve this goal?

Enter VFD networking

Networking, or network integration, refers to the process of connecting to devices in the building via a communication protocol for the purpose of monitoring and/or control. The BMS of the past had one pair of wires for each signal, which typically used either contact closures, 0-10 V dc, or 4-20 mA signals to either control or monitor a single piece of data. Depending on the installation, these discrete inputs and outputs may or may not have been physically connected to the device performing the control function. Networking allows direct communication between the controlling device and the VFD. Controls contractors typically factor in the count of monitoring and control points in their price. In an office building where there could easily be a few dozen VFDs, the cost of adding a hard-wired point to each VFD can quickly result in significant costs. Due to these limitations, VFD control and monitoring traditionally were limited to start commands, status feedback, speed command, and possibly a general alarm status. By networking VFDs, far more data is available. Information, such as motor frequency feedback, specific alarms, and whether the VFD is in “hand,” “auto,” or “local off” modes can be monitored and displayed by the BMS. While this data has long been available via networking, one wonders if reasons other than cost prevented this functionality from being used.

Figure 2: Monitoring data from variable frequency drives (VFDs) is only part of the solution. Trending at appropriate sampling rates to correctly see equipment and system response in the event of an alarm is critical to performing a root-cause analysis. Courtesy: Page

Among others, the number of communication protocols on the market was dizzying. Each controls manufacturer had their own proprietary protocol, while none of the manufacturers were particularly strongly integrating with other protocols. This made coordination of communications between VFD manufacturers and controls vendors a headache with which few were willing to deal. In 2004, ASHRAE issued ASHRAE 135-2016: A Data Communication Protocol for Building Automation and Control Networks (BACnet Standard) in an attempt to create a standard communication protocol for the HVAC industry to aid in the process of sorting through the muck. While the process has taken time, and the protocol has its drawbacks, the underlying goal of creating a common language that allows multiple manufacturers’ equipment to communicate with one another has been achieved. It is important to note that past reliability issues with equipment networking have been substantially reduced through time and experience in installing communication networks as well as the BTL (BACnet Testing Laboratories) Certification process, which certifies individual devices for compliance with the BACnet communication standard. There are many additional challenges associated with the physical network infrastructure, which are beyond the scope of this article, though many are tied back to the logistics of how communications networks are physically installed.

Another major roadblock to increased integration is the manner in which BMS contractors manage their pricing structures and front end. With the exception of Niagara, most BMS server databases and graphics development platforms are proprietary. Many owners and engineers also are unaware of the difference in the efforts required to add a networked point versus a physical point. After the device has been networked, the process of adding additional points, provided the equipment is already configured to broadcast the desired point, typically takes a matter of minutes (multistate points, such as equipment status, may take longer due to the information available). Compare this to ensuring a physical input/output (I/O) point exists on a controller, having an electrician run conduit (depending on the project) and wire to the device, and making terminations, and the relative cost should be clear. However, many controls vendors still charge the same for networked points as they do for physical points. It is important to note, however, that in both cases there will be cost for adding the points to graphical interfaces, adding alarm conditions, and adding trends. Though BACnet allows controllers and equipment from different manufacturers to talk to one another, the proprietary nature of most server databases and graphics platforms puts most owners at the mercy of their installed vendors’ pricing to make those new points visible to the operators.

The value of networking VFDs

Networking allows the collection, storage, and later review of a broader spectrum of data by multiple individuals for the purposes of troubleshooting problems, identifying potential issues before they occur, and enhancing delivery of service to the occupants of the building (see Figure 1). For example, power- and energy-consumption data points are available through most VFDs today. A key tenet for high-performance buildings is knowing where power is consumed. By monitoring these VFD parameters, users know what these systems consume. However, more important is the ability to trend this data. If users trend the power consumption of an air handler serving an open-office plan, they may see that the consumption is greater this month than last. This could be due to several factors including occupant count and outdoor ambient temperature. However, it could be due to the air handling unit filters beginning to load, increasing the pressure on the fan. By monitoring and trending this power-consumption data, operators are able to identify when equipment and systems are operating at other than peak efficiency. By reacting to this information, operators can better maintain the systems for optimum energy efficiency and reduce energy costs.

While able to be monitored via a hard-wired input, the status of the VFDs hand/off/auto switch can aid tremendously in troubleshooting drive issues. On multiple occasion, hot or cold calls have been logged on a given system, indicating the system was not delivering sufficient airflow. Upon looking at the VFD, it was noted that someone placed the VFD into hand operation, locking the VFD at a given speed. In many cases, this is done because of issues with the control loop, or someone simply forgot to place it back in auto mode after performing maintenance on the system. However, monitoring this point allows visibility to the mismatched operation and allows the team to determine the reason. Whether left in hand mode accidentally, or due to a faulty control loop, degraded service to the occupants and an impact on energy use are the results.

Another data point available in most VFDs today is motor-speed feedback. A common issue observed when reviewing the operation of a poorly controlled or excessively slow hydronic pumping system is a mismatch between the VFD setup and control programming that causes the systems to fight each other. VFDs have settings that affect the maximum acceleration and deceleration rates of the motor. No matter what the BMS is commanding the VFD to do, the VFD will not accelerate or decelerate the motor faster than these ramp settings. The BMS pump speed control loop assumes that the pump speed is changing effectively in sync with the change in commanded speed. If the VFD ramp rates are set slower than the BMS control loop is configured, the VFD will lag behind the BMS command, causing a hunting issue. This is most commonly observed when starting and stopping pumps and is exacerbated on critical chiller plants, which typically require very fast response to disturbances in the system. This issue can be easily diagnosed by monitoring the motor speed feedback relative to the commanded speed. Correcting the issue also can be easily addressed by ensuring the VFD ramp rates are set faster than the BMS control loop rate. A nice benefit to motor speed monitoring is a second means for determining if a VFD is placed in hand mode versus auto mode.

Using trend data

When it comes to system monitoring, one of the most critical uses of trended data is forensic analysis (see Figure 2). When a system has failed, time is spent looking at the data available to determine a root cause. Sometimes there is sufficient data; but all too often, the data is lacking. This usually is due to a combination of insufficient channels monitored and insufficient trend rates. Historically, the list of channels to be trended is left to the BMS contractor and the owner’s operators. Depending on the contractor on the project team and the experience of the operators, the project might end up with no trends other than temperatures and flows—or it may end up with a trend on every point regardless of its benefit to the operators.

Traditionally, unless otherwise specified, HVAC trends of analog values are stored once every 15 minutes or so. While the actual intervals in use may vary, the order of magnitude is similar. However, in the event that a pump fails on a chiller plant on an overcurrent trip, trend intervals in the 15-minute range are insufficient to identify the sequence of events (see Figure 3). This leaves the user to ask the following questions:

  • What was the current draw at the time of the trip?
  • What was the ac input voltage at the time of the trip?
  • When the pump failed and the lag started, how long did it take for the flow in the plant to recover?
  • How quickly did the VFD accelerate from “off” to the commanded speed?
  • What was the lowest flow recorded?
  • What was the impact on the loop temperature?

If the temperatures, flows, and pump speeds are trended at only 15-minute intervals, there is no data for review. Trend rates can be addressed easily in a new BMS due to the increased throughput of their data infrastructures and larger hard drives. A modern-technology BMS is capable of running multiple trends on a trend. One trend can be used for long-term monitoring using 15-minute (or similar) intervals, while the other is passed through a first-in/first-out buffer at 1- to 5-second intervals and saved to the historian only when an alarm condition occurs. Though this functionality is not native in today’s BMS—custom programming usually is required—it is available. Channel monitoring and trending comes down to specification and networking.

Equipment tripping offline due to fault conditions is a fact of life in this business. Most equipment, such as VFDs with solid-state controllers, have the ability to individually identify different fault conditions. However, if the VFD monitoring is performed via a hard-wired input, all that is logged and trended in the BMS is typically a general fault condition. While most VFDs have the ability to individually program relay contacts to actuate on specific alarms, monitoring more than one requires additional pairs of signal wires. If the VFD is networked, however, literally dozens of discrete alarms can be monitored, logged, and trended, providing end users with a clear, consolidated history of alarm events with the VFD. Additionally, this functionality allows multiple alarms occurring simultaneously to be seen and prioritized, which a general alarm does not provide. Again, to ensure a high quality of service to the occupants, this type of data arms the building operations team with the information necessary to quickly restore the system to proper operation (see Figure 4).

More networking options

In addition to manual data review and analysis, numerous products are coming to market aimed at the high-performance building market. In many cases, these products are black box add-ins to the BMS that work to collect, analyze, and optimize HVAC system operations based on building-specific needs and equipment. For these systems to operate to their fullest potential, substantial data on motor loads and consumption typically is required. Every major BMS vendor has some form of data analytics that can be added or enabled in the system to leverage the data collected. It is important to coordinate with the owner and permitted bidders during the design phase to understand what the owner wants and what the vendors can provide. Because every vendor has a different catch-phrase to describe a given solution, it is typically best in a competitive bidding environment to specify analytics packages with performance-based language that align with owner needs. Due to the rapidly evolving offerings, it also is recommended to require a live demonstration during the bid-evaluation phase, if not prior.

Many more useful pieces of data are available over the network interface to most HVAC VFDs today. These range from simple monitoring points, such as voltage and current draw, to complex analytical points, such as total harmonic distortion and power factor, when specified as options on the VFD, which become very 

important for large motors. Relative to the traditional monitoring and control used for VFDs, the available information is almost limitless, but at what cost? When discussing this topic with multiple control vendors in various regions of the country, the answer tends to be fairly consistent: Networked points cost approximately 10% of hard-wired points. Of course, this can vary greatly depending on what is done with the point and whether integration to the device already exists. 

Figure 3: This trend shows the importance of appropriate sampling intervals for trends. If these trends were sampled at only 15-minute intervals, the severity of the control problem would not be clear. Courtesy: Page

But in general, points in addition to the first for a given networked device are available at a fraction of the cost of the hard-wired point. This is due to both the cabling requirements for hard-wired points as well as possible need for a field device to generate the data to the BMS. For example, a current switch. Another advantage to networking VFDs—or other equipment, for that matter—is that points can be added to a device easily and relatively inexpensively when the initial network connection is established. During the life cycle of a building, different information becomes important depending on the staff and the goals of the operations team. Instead of bringing contractors into the space and possibly disturbing occupants to add conduit and wire, some programming at the BMS allows the new points to be integrated into the system.

Justifying VFD networking

If networking a VFD is so cost-effective (relatively speaking), and so much data is available, why do we use hard-wired points at all? In most cases, it comes down to one thing: the criticality of the system served by the VFD. It is important to note that many VFDs still use BACnet MS/TP (serial) communications. As a result, powering down a device on the network, which may be completely unrelated to the system, can affect the speed of communication with the remaining devices. Because VFDs are operating one part of a control loop, with the commanded speed typically coming from the DDC controller, this can lead to a lag in the command reaching the VFD. This results in a reduction in the quality of control for the system, which may or may not be noticed by the building occupants. In some processes, such as data centers, semiconductors, medical surgical facilities, and others, this potential risk is not acceptable. In these cases, a typical VFD configuration may include hard-wired signaling for the basic and critical control functions, such as start command, speed command, status, and general alarm (does this data set sound familiar?), and networking for the monitoring points, such as energy consumption, discrete alarming, harmonics, and so on. This type of hybrid approach maintains the direct connection between the DDC controller and VFD, independent of outside influences, for all critical operations, but still provides a cost-effective means for monitoring the data that allows the team to maintain optimized operation of the system.

Another available option is to have separate subnetworks for devices on a common system. Depending on the project, the geographical layout of the hardware, and the number of potential subnetworks, it may be acceptable to install a handful of subnetworks that ensure only related equipment is on a given communication network. 

Figure 4: An important aspect of specifying networked data is making it usable to the end user. In many cases, data (such as “drive fault code”) are transmitted as multistate values with a lookup table in the user manual to decipher the meaning of each value. Simply displaying the raw BACnet value of the multistate value does not help an operator. However, translating that value based on the table in the variable frequency drive (VFD) manual—in this case, to “dc undervoltage”—provides concise and meaningful data for the operator. Courtesy: Page

This could be useful for many reasons including limiting the data traffic on given networks; compartmentalizing systems for facilities that may see frequent renovations, such as lease spaces; and fault tolerance for critical systems. If choosing this approach, however, it is critical to follow up during submittal review and installation to ensure that this networking approach is maintained by the contractor.

As mentioned earlier, data throughput is much less of a problem today than in years past. But much like any technology, users will find a way to use all the capacity something provides them. During the design of any BMS that will use networking of equipment, it is important to have either the engineer or contractors include a network traffic analysis to confirm sufficient infrastructure exists to support the desired data throughput. To properly conduct this analysis, detailed specifications should be provided indicating the points intended for monitoring as well as the trending rates of the various points. Likewise, a minimum duration of data storage, rather than a particular storage size, should be specified to confirm sufficient data will be readily available to the building operations team for future use. If the data is not readily available, it will not get used regularly. Thus, the end goal of the networking and data collection will be lost.

From an HVAC perspective, the end goal for high-performance buildings is to provide a high-quality indoor environment with high reliability and minimal energy consumption. Recent changes in the market approach have complicated this goal by reducing head count or outsourcing building operations teams, thereby forcing this service to be delivered with less site experience and manpower. One of the most efficient ways to empower the remaining staff is through the efficient collection and use of operational data, whether directly or through building automation. Depending on the building use, either complete networking or a hybrid approach can bring substantial data to the hands of the building operations team to detect downward trends in building performance, quickly identify root causes of failures, and educate new staff on historical building operation.


John Gross is associate principal/senior project manager and senior mechanical engineer at Page in Houston. With 13 years of experience in data center, green building, and large chiller plant design and commissioning, Gross is Page’s lead DDC controls and forensic analysis engineer.