Sometimes improved power quality and reliability aren't the only justifications for installing backup power

Almost every engineer and facilities manager understands the need for improved power quality in mission-critical applications. IT equipment has generally met or exceeded ITIC — formerly CBEMA — voltage tolerance requirements for more than a decade, but manufacturing loads often lack such stringent voltage tolerance levels. In fact, many of these loads require much better voltage regulation, and are less tolerant of short-term voltage transients. At the same time, they may be highly nonlinear or apply much larger step loads than data processing equipment. This greater sensitivity to voltage variations and waveform distortion coupled with harsher load characteristics reduces power quality in industrial applications. These characteristics also make processes more susceptible to poor power quality.

Given this situation, it seems surprising that manufacturers don't widely use UPS systems in their operations. But the fact is few manufacturing operations have critical power requirements, making it difficult to justify the cost of a UPS system. In most cases, the effects of poor power quality are subtle and incremental rather than blatant and catastrophic. In addition, manufacturers often include the bulk of power quality costs in the standard cost of products, and only unusual power quality events like extended power outages show up as cost variances. In fact, until recently, manufacturing companies had little choice other than accepting the effects of power fluctuations. These companies used the built-in allowances for capacity, scrap rates, overtime, and rework costs to cover less-than-optimum power quality. Therefore, the biggest part of power quality cost may be unknown to manufacturing management. Management may discover the cost of poor power quality only after facing in increasing pressures to reduce production costs and increase efficiency, forcing them to make deeper analyses of every aspect of the manufacturing operation.

Further hampering the wider use of conventional UPS systems has been the substantial engineering risk involved in applying them to manufacturing applications. Significant compatibility issues have affected their use with manufacturing loads that are nonlinear and, often, very specialized. These frequently include large variable speed DC drives or other high-power phase-controlled SCR devices, motors, and other cycling loads.

Battery-based UPS systems.

In the past, battery-based UPS systems haven't been able to meet the financial criteria required to justify capital expenditures in most manufacturing environments for several reasons. First the total cost of the UPS installation, including facility modifications to create a suitable environment for the batteries, has been too high, even when the cost of UPS equipment alone is well within required capital cost limits.

Second, continuing UPS maintenance costs, including those resulting from short battery life and the continuing cost of scheduled and unscheduled battery replacements, significantly increase annual operating costs of the process under protection (Sidebar “The Cost of Replacing UPS Batteries” below). In addition, utility costs increase due to electrical losses of the UPS equipment.

Finally, the possibility of process interruptions due to UPS system misapplication or failure (risk) offsets potential financial benefits of conventional UPS systems.

Today, the valve-regulated lead acid (VRLA) battery is commonly used at all UPS power levels. It's also the predominant battery used in manufacturing operations with a UPS system because it costs much less and requires much less space than conventional vented lead acid batteries (wet cells). But batteries, particularly VRLA batteries, are recognized as the UPS system component with the highest failure rate.

A paper titled “Increasing UPS Battery Life: Main Failure Modes, Charging, and Monitoring Solutions,” written by members of the R&D Advanced Technologies Team of MGE UPS Systems, details a study conducted on VRLA battery failures. Two methods were used. The first was a theoretical analysis of failure mode effects (FMECA) on a 10kVA, single-phase UPS, where battery behavior was evaluated in three life phases: charge, discharge, and charge maintenance. The second was a practical analysis of after-sales maintenance databases. The results were compared “for validation of main failure modes.” Using a database of 8,000 records, the study also highlighted the average elapsed time before the first battery maintenance operation, which included partial or total replacement of the battery. Fig. 1 presents the findings of the study.

Based on this failure data, you could expect a 240-cell battery system, which is the most common configuration for large UPS systems, to average five cell failures in the first year (2% of 240 cells), nine in the second (4%), 48 in the third (20%), and 84 in the fourth (35%). Since VRLA batteries predominantly fail open-circuit, any single-cell failure results in failure of the entire battery system. So the mean-time-between-failure (MTBF) of the entire battery system is measured in months or weeks rather than years (Sidebar “Calculating the MTBF of Battery Strings” above). Therefore, eliminating batteries improves overall reliability and eliminates the continuing costs associated with battery maintenance and replacement.

Flywheel-based UPS systems.

In addition to the elimination of batteries, UPS systems designed specifically to work with flywheels can have operating characteristics that make them better suited and more reliable for use in manufacturing operations. For example, UPS systems with integrated flywheel energy storage have a higher DC bus voltage and lower impedance, which allows these devices to support DC drives and other commutating loads while maintaining a better output voltage waveform. Also, flywheels can deliver higher peak currents than the short run-time batteries typically installed in industrial operations; longer run-time batteries make UPS systems cost prohibitive. Therefore, without significant over-sizing, flywheel UPS systems are better able to supply the difficult loads often encountered in manufacturing operations. This reduces capital cost and makes UPS protection financially attractive to more manufacturing operations.

Most flywheel UPS units use a line interactive design that makes them more efficient than conventional double-conversion battery UPS systems, thereby lowering utility costs that can result from UPS efficiency losses. UPS systems designed with integrated flywheel energy storage typically achieve efficiencies of 97%, compared to 93% or 94% for conventional battery-based UPS units.

In addition to their electrical characteristics, the reduced space requirement and wider operating temperature range of flywheel UPS systems makes them easier to retrofit and less expensive to install.

Cost modeling.

Evaluating the financial benefit of UPS protection and the relative benefits of alternative UPS solutions can be a difficult and time-consuming task. It requires a model that includes many variables, some of which may not be obvious without an in-depth understanding of UPS systems and their applications. It's also helpful to develop typical parameter values to use as a starting point in the model. This allows you to make preliminary assessments before dedicating substantial resources to project engineering and estimating.

A simple model was developed in 1999 to analyze the return on investment (ROI) and payback for a 1,000kVA conventional battery-based UPS and a 1,000kVA flywheel-based UPS over a 10-year period. The model included cost inputs for total initial capital investment, annual power quality savings, electricity cost, and total maintenance cost. It also assumed that the continuing costs remain constant over time, after adjustment for inflation.

The 1999 analysis used two conventional battery-based static UPS modules in parallel as the comparison to a single flywheel UPS. This was a reasonable comparison at the time because conventional UPS modules with ratings greater than 750kVA weren't generally available. Now, however, conventional static UPS modules are available in ratings as high as 1,000kVA. Further, the earlier model didn't contain sufficient detail to give insight into all of the costs that accompany the installation and ownership of a UPS system.

The financial analysis methods used today are based on a new model that accepts detailed cost estimates to generate monthly cash flows for 10 years and calculates the resulting ROI and investment payback period. The new model also allows much more accurate analysis of costs associated with periodic full replacement of battery strings and interim partial string replacements.

This article also draws upon financial analyses that use a 900kVA flywheel UPS system and a 1,000kVA conventional static UPS equipped with about 6 min of batteries. The ROI and payback results for the two UPS systems are shown in Fig. 2 and Fig. 3. As you can see, the flywheel UPS shows significantly higher ROI and shorter payback for the base case and throughout the -50% to +100% sensitivity range of each variable.

The same four variables included in the 1999 study were then graphed for a flywheel-type UPS and for a battery-type UPS. An additional variable for battery life was also included for the latter. The sensitivity line for battery life shows the effect of a battery with a longer or shorter life, not simply the effect of changing a given battery at shorter or longer intervals. Each curve indicates the effect on ROI or payback that results from a percentage change in one variable — as shown on the horizontal axis — when all of the other variables remain constant at base case values. The base case for each UPS occurs where the sensitivity lines for that UPS intersect. Solid lines denote flywheel-type UPS variables while dashed lines denote battery-type UPS variables. Each of the variables graphed — except battery life — is a combination of several inputs to the model. Installed cost of the battery UPS, for example, is the total of inputs for the UPS module and accessories, maintenance bypass, batteries, battery safety systems, and installation.

Capital investment and operating costs.

The steep curves for the total installed cost of the UPS systems in Fig. 2 and Fig. 3 show that capital investment is extremely important in determining the financial viability of UPS systems in industrial applications. A 10% increase in total installed cost decreases the ROI by 7% and 5% for flywheel- and battery-based UPS systems, respectively, assuming all other factors remain unchanged. Correspondingly, this same increase in total installed cost increases the payback period by about two months for either system.

The total installed cost figures don't include any oversizing of the UPS systems to accommodate industrial loads. The financial effect of oversizing a battery-based UPS by 25% results in an increase in capital cost by close to 30%, which reduces the ROI accordingly. The oversizing also increases the monthly payback time by the same percentage (Fig. 4 and Fig. 5). At the power levels analyzed, oversizing is costly for conventional battery UPS systems because module capacity limitations require you to parallel multiple modules to increase capacity beyond 1,100kVA. You're less likely to need oversizing with a flywheel UPS system.

Operating costs affect ROI and payback much less than the installed cost of the UPS does, as indicated by the relatively flat curves for operating cost variables in Fig. 2 and Fig. 3. However, UPS operating costs will have a significant effect on annual plant overhead budgets. Annual operating costs for a flywheel UPS, in current dollars, will amount to about $29,000, assuming electricity cost of $0.07 per kW-hr. Battery UPS yearly operating costs vary between $51,000 in years when battery replacements are minimal, to about $105,000 in years when you need full battery system replacements.

A conventional, double-conversion, battery-based UPS in this size range operates at around 94% efficiency under optimum conditions. At conservative loading levels and $0.07 per kW-hr. average utility rate, the UPS losses will increase utility cost by about $38,000 for battery UPS systems.

As noted before, a 900kVA line-interactive flywheel UPS operates at 97% efficiency. Utility cost increase at the same loading level is about $20,000. Utility cost calculations include electricity consumed by air conditioning to remove heat created by the UPS system at 1.8kW per ton of HVAC.

For the base case, electricity consumed by each UPS reduces ROI by 9% and 18% for flywheel UPS and battery UPS, respectively, compared with the ROI achieved if each UPS system were 100% efficient. Because utility consumption has twice the effect on ROI for the battery UPS as it does for the flywheel UPS, variation of UPS efficiency from the base case also changes the ROI for the battery UPS twice as much as it does for the flywheel UPS. Base case efficiency losses increase payback periods by two months for the flywheel UPS and five months for the battery UPS.


Reducing the risks associated with designing and specifying a UPS project is one way in which flywheel UPS systems increase predictability of the overall power system. Another aspect of predictability is predicting equipment failures.

VRLA battery system failures aren't easily predicted without installing a battery monitoring system that may cost as much as the battery system itself. In contrast, flywheels, like many other mechanical systems, give ample warning of failure. Temperatures rise, vibration increases, and operating parameters, in general, start to change gradually rather than failing suddenly. Flywheel UPS systems constantly monitor all of these parameters, allowing trending and predictive analysis.

This capability not only avoids sudden flywheel UPS failures, it also allows consistent budgeting of UPS maintenance costs. Large unbudgeted costs are always an undesirable occurrence for manufacturing management.

More importantly than the fact that flywheel-based UPS systems yield higher ROIs than conventional battery-based UPS systems in large manufacturing operations, they allow more facilities to achieve project payback in less than two years, which is a widely applied criterion in determining the financial viability of manufacturing capital investments.

Walter is director of product marketing for Active Power, Inc., Austin, Texas.

Sidebar: Calculating the MTBF of UPS Battery Strings

You can't calculate the MTBF of an entire UPS battery string by dividing the mean life of a cell by the number of cells in series (46 months ÷ 240 = 0.19 months) because the data from the MGE UPS Systems study indicates that the probability of failure of individual cells is far from constant over time. The MTBF will depend on the age distribution of cells in the UPS and therefore will be longer when the battery string is new and shorter once it's more than about three years old.

Sidebar: The Cost of Replacing UPS Batteries

For a UPS system in the 1,000kVA (800kW) range, the total cost of all battery replacements for a battery UPS over a 10-year period is estimated at $250,000, based on the failure rate data from a study conducted by MGE UPS Systems and detailed in a paper titled “Increasing UPS Battery Life: Main Failure Modes, Charging, and Monitoring Solutions.” The results of this battery study support the practice of replacing VRLA batteries at the end of the third year of service.

Today's replacement cost in terms of labor and materials of a 6-min VRLA battery system with a 10-year pro-rated warranty is about $60,000. You'll also need multiple interim cell replacements every year between replacements of the complete battery system.

Service call charges can amount to $1,000 or more per trip to the customer site. Since most UPS maintenance contracts include only scheduled preventive maintenance for batteries, all service calls for battery replacement or other battery remediation are a sizable hidden cost of UPS ownership. The cost of interim replacements varies greatly from year to year but will average about $5,000 annually over the 10-year period.