828 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 17, NO. 3, JULY 2002 The Design of Power Acceptability Curves John Kyei, Student Member, IEEE, Raja Ayyanar, Member, IEEE, Gerald Heydt, Fellow, IEEE, Rao Thallam, Fellow, IEEE, and John Blevins, Member, IEEE Abstract—The concept of “standards” is introduced for the de- sign of power acceptability curves. The power acceptability curves are aides in the determination of whether the supply voltage to a load is acceptable for the maintenance of a load process. The con- struction of the well-known Computer Business Equipment Man- ufacturing Association (CBEMA) power acceptability curve is dis- cussed, and issues of 3-phase and rotating loads are discussed. Index Terms—Computer Business Equipment Manufacturing Association (CBEMA) curve, power acceptability curves, power quality, voltage sags, voltage sensitivity, voltage vulnerability. I. POWER ACCEPTABILITY M ANY power quality indexes relate to steady-state phe- nomena, and relatively few relate to momentary events. However, many power quality engineers feel that bus voltage sags, a natural consequence of a highly interconnected transmis- sion system, may be the most important type of power quality degradation, and a useful measure of the severity of these events is desirable. One such metric is the power acceptability curve (or voltage sensitivity or voltage vulnerability curve) which is a graphic metric of the severity of bus voltage sags plotted versus the duration of these events. The best known of these graphical metrics is the Computer Business Equipment Manufacturing Association (CBEMA) curve which is a graphic depicting the severity of a distribution bus voltage sag versus its dura- tion . The plane is a two-dimensional (2-D) space in which the line represents the case that distribu- tion voltage is at rated value, and the half-plane is the bus voltage sag region. Overvoltage and undervoltage events of very minimal impact (small ) are considered “acceptable” in the sense that loads are not disrupted; furthermore, very short duration events (small ) are considered acceptable. Thus, the plane is divided into acceptable and unacceptable re- gions. Fig. 1 shows the CBEMA power acceptability curve. The CBEMA curve depicted in Fig. 1 has indicated as a percent of rated voltage, and shown on a logarithmic scale in seconds. Other formats of the curve may entail different voltage scales and linear scales (possibly depicted in cycles). The CBEMA curve has an upper limb locus that depicts the boundary of acceptable/unacceptable operation for overvolt- ages. Because many power quality engineers believe that bus voltage sags are particularly important, the focus of this paper is on the undervoltage (lower) limb in power acceptability curves. Manuscript received May 23, 2001. This work was supported by the Power Systems Engineering Research Center (PSerc) and SRP. J. Kyei, R. Ayyanar, and G. Heydt are with Arizona State University, Tempe, AZ 85287 USA. R. Thallam and J. Blevins are with the Salt River Project, Phoenix, AZ 85072 USA. Publisher Item Identifier S 0885-8977(02)05922-8. Fig. 1. CBEMA power acceptability curve. References [4]–[6] discuss a fuzzy logic alternative to assess voltage—load sensitivity, testing of loads to CBEMA standards, and computer performance during voltage sags re- spectively. Bollen and Zhang [7] have discussed a classification system of voltage sags and their effects. Ride-through issues for adjustable speed drives appear in [8]. It is evident that power acceptability curves have frailties in design and application. For example, very short duration events (e.g., less than a cycle in duration) have an ambiguity in the sense that the duration of the event may be difficult to identify, and the point-on-wave of the disturbance may have significant impact on the load. Point-on-wave information is not depicted in the plane. Furthermore, the 3-phase implications of a power acceptability curve as indicated above are not clear: should one utilize phase information in the plane, or the positive sequence of the distribution voltage? Or is the graph ba- sically a single-phase representation? Another commonly asked question relates to the equation of the loci shown in Fig. 1. The CBEMA curve was developed from experimental and historical data; that is, cases of load disruption of mainframe computers were plotted in the plane, and a separator was developed to identify the acceptable and unacceptable regions. However, what is the scientific basis of this separator, and does the sepa- rator apply to all load types? These issues are addressed below in connection with an organized design of power acceptability curves. Although the CBEMA curve was singled out above as a power acceptability curve, there are other similar curves [2], [3]. [1] contains a discussion of the foibles of power accept- ability curves. Difficulties in the measurement of compliance with the CBEMA curve, and widening of the acceptable region 0885-8977/02$17.00 © 2002 IEEE