AMERICAN PUBLIC TRANSIT ASSOCIATION 2003 RAIL TRANSIT CONFERENCE. Cable Rating Considerations for Direct Current Traction Power Systems

Transcription

1 AMERICAN PUBLIC TRANSIT ASSOCIATION 2003 RAIL TRANSIT CONFERENCE June 9, 2003 Rating Considerations for R. W. Benjamin Stell. P.E Manager of Power Systems The HNTB Companies 1 Burlington Woods Burlington, Massachusetts (781) (781) (Fax) bstell@hntb.com ( ) Pre-Conference Draft Paper

2 Rating Considerations for ABSTRACT circuits in direct current (dc) traction power systems typically conduct load currents that vary greatly in magnitude and duration. These load variations also change significantly with time of day in reasonably predictable patterns. In addition, these cables are often installed in duct banks that are either larger or configured differently than those included in industry cable "ampacity" (rating) tables. Commonly available ampacity tables in North America were not developed for application to dc traction power cable circuits, and can provide overly conservative results when so applied. These reasons contribute to the difficulty of selecting the most cost-effective cable size and quantity for a traction application, or for determining the conditions under which existing cables will be overloaded. This paper will review concepts, common assumptions, and industry standards associated with the rating of underground power cables as they apply to dc traction power application. Steady state and transient cable rating analyses, as well as available cable ampacity tables and the assumptions on which they are based will be discussed. The differences in approach between the rating of standard power cables and cables for dc traction applications will be distinguished, and guidelines provided. The importance of using commercially available computer software for analyzing both the steady state and transient current-carrying capacity of typical traction cable configurations will also be confirmed. INTRODUCTION The amount of current that a power cable can safely conduct under a specific set of conditions is known as its current-carrying capacity, also termed its "rating", or just "ampacity". The process of analytically determining the ampacity of multiple underground power cables is a challenging technical exercise involving the application of heat transfer theory. Considerable research performed during the past 100 years has resulted in the development of sophisticated techniques for calculating cable ampacity [1]. These methods are exemplified by the International Electrotechnical Commission (IEC) Standards 287 and 853, and by their North American precursor, the Neher-McGrath Model [2]. To avoid the complex calculations associated with these methods, many engineers refer to published tables to obtain cable ampacities. The available tables, however, are based on installation and material assumptions which may not be readily apparent or even applicable to a given installation, particularly a dc traction power system duct bank. In addition, the available ampacity tables for underground installations were developed for application to balanced three-phase alternating current (ac) systems, and for a limited number of configurations particular to utility and industrial ac power systems. CHARACTERISTICS OF DC TRACTION POWER CABLE INSTALLATIONS North American dc traction power distribution systems typically utilize multiple large diameter, unshielded, single core copper cables insulated for 1000 or 2000 Volts. s known as positive feeders connect the 600 to 750 Vdc nominal substation dc cathode buses to the right-of-way contact system. Negative return cables connect the running rails or, in the case of electric trolleybus, the negative contact system, to the substation negative (anode) bus. The positive and negative cables may be installed in one large ductbank, or in separate ductbanks. In either case, the positive and negative cables from/to the substations are typically installed in concrete-encased duct banks ranging from at least 8 ducts for light rail systems to as many as 36 ducts for some heavy rail systems. The positive and negative cables carry widely varying load currents on a moment-to-moment basis as well as on an equivalent rms basis over a 24 hour day (the equivalent rms being the steady current equivalent of the instantaneous load variations over defined time periods). On an instantaneous basis, cable current may vary by factors of 200 or more between train starts. Over a normal 24 hour period, these cables will typically sustain two periods of peak current during the morning and evening rush hours. Before, in between, and after the rush Page 1 of 8

3 Rating Considerations for hours the cable load will be substantially less and, in many cases, the cables will have essentially no load in the early morning hours for systems that do not operate 24 hour revenue service. Periods of intense "catch-up" service may occur after equipment failures or operational difficulties resulting in increased cable loading in the affected areas; the duration of these events is limited by the amount of available rolling stock, however, and thus they rarely exceed 30 to 45 minutes. CABLE RATING CONSIDERATIONS Maximum Operating Temperatures The object of the cable rating process is to ensure that cable components are not damaged by high operating temperatures. Temperature rise in cables is caused by the electrical losses that occur in a cable as a result of current flow. For cables carrying alternating current, these losses include resistive "copper" losses in the cable core as well as additional heating due to skin effect, proximity effect, sheath and shield losses, and dielectric losses. For cables carrying direct current, however, only the "copper" losses contribute to temperature rise. Modern traction power cables are designed to be operated continuously at a maximum temperature of 90 Degrees C. In accordance with North American cable standards (ICEA and IEEE), these cables are also designed to withstand an emergency overload temperature of 130 Degrees C, provided this temperature is not sustained for more than 100 hours in any one year, or 500 hours during the lifetime of the cable. Rms Current The procedures for calculating cable ampacity may be divided into two major categories, steady state and transient. Steady state cable ampacities are obtained by applying a single continuous value of current to one or more cables for a sufficient period of time to enable cable temperature to stabilize, or reach a steady state. The actual time required to heat the cable to its normal temperature limit, which is a function of the cable's size and the "thermal capacitance" of the installation environment, is not considered. Transient ratings involve the calculation of cable ampacities and operating temperatures as a function of time. Ampacity tables provide steady state cable ratings only, and are designed to work with essentially steady, non-varying values of current. For the typical ac circuits to which these tables are intended to be applied, this steady value would be the root mean square, or rms value. However, if a recording ammeter is installed on a typical LRT or heavy rail feeder circuit, a widely varying waveform will be observed. The same waveform, in digital sampled form, can be reproduced by a competent traction power load flow simulation program. This waveform will approximately repeat itself with the headway cycle of the vehicles being powered by the cable. In other words, the cable current waveform will be periodic with the scheduled headway. This waveform must be converted into an equivalent constant value of current before it can be compared with ampacity tables or used in steady state ampacity calculations. A load flow simulation program normally performs the required root mean square calculation that accomplishes the conversion. A root mean square calculation produces an equivalent constant value of current from any periodic timevarying waveform. The rms value of a periodic current waveform is defined as the equivalent constant direct current which would cause the same average heating in a resistive element, such as a length of cable. The rms current calculation, when properly applied to a current waveform, produces a current value which can be compared with those found in published cable ampacity tables. It is important to stress that the rms calculation is only defined for a periodic waveform, and is only valid over an integer multiple of the waveform period. As noted above, for traction power systems the current flow in cables is typically periodic with vehicle traffic (that is, with vehicle type, consist, and headway). If different vehicle consists or schedules are interspersed, then the waveform period will become the least common multiple of all the headways operating in the area of interest. The choice of any other time period could result in calculated rms currents that are higher or lower than the actual rms current. Page 2 of 8

4 The Load Factor Rating Considerations for Load factor is defined as the average load over a specific time interval divided by the peak load occurring during the same interval. For cable rating purposes, the time interval utilized is typically a 24 hour day, which results in a quantity termed "daily load factor". The daily average and peak loads are not instantaneous values but average demands, and must be expressed in the same units [3]. Hence the daily load factor is a dimensionless quantity, and is less than one (100%) for typical cable circuits (in other words, for cables powering loads that are not constant for 24 consecutive hours). The concept of daily load factor is employed in the traditional cable rating process to approximate the effect of varying load current on cable operating temperature. Large diameter power cables installed underground have considerable "thermal capacitance", and therefore may not heat up quickly under normal loading. If a large diameter cable rated for short peak load periods carries significantly less load at other times, it will not reach its 90 Degrees C rated ampacity during the peak load periods. It will also have an opportunity to cool off between peak load periods. For this reason, a cable with a load factor of less than 100% will have a higher ampacity than a cable operating at 100% load factor. With the exception of the National Electrical Code Article 310 tables, published tables for underground installations list ampacity values for one or more load factors. Available computer software packages for calculating cable ampacities also permit the use of daily load factor in steady state ampacity calculations. The determination of daily load factor for a traction power system can be made in several ways. It can be derived from averaged SCADA system cable load data, provided the data points are sufficiently frequent. It can also be approximated from daily system operations timetables, under the assumption that cable load variation is reasonably proportional to the variation in the number of trains per hour passing by the affected substation. It can also be determined by computer simulation of a typical daily transit system schedule. Application of these methods results in daily load factors for typical transit systems that rarely exceed 75%, even for rapid transit systems with 24 hour revenue service. However, load factors will not be identical for all cable circuits and, strictly speaking, they should be calculated separately for each cable circuit. In addition, as will be illustrated below, this traditional approximation is no longer necessary now that sophisticated cable ampacity simulation programs have become commercially available. REVIEW OF AVAILABLE CABLE AMPACITY TABLES ICEA Standard P /IEEE S-135 The current version of this standard was published in 1962 in two volumes, and reprinted with minor corrections in 1966 as a single volume known as the ICEA "Black Book". It was the result of a joint effort between the Insulated Engineers Association (ICEA, formerly IPCEA), and the Institute of Electrical and Electronic Engineers (IEEE) Insulated Conductors Committee. As with all ampacity tables, the ampacities listed are for the "limiting cable" in the installation; this is the "hottest" cable of all the cables in the duct bank. Even though all the remaining cables normally operate at lower temperatures, all cables in the duct bank are assigned the ampacity of the limiting cable. All cables in the duct bank are assumed to be of identical construction, and equally loaded. ICEA P contains ampacity tables for a wide variety of cable types, voltage levels, installation configurations, and load factors. The ampacities are calculated using a modified version of the Neher- McGrath Model. The tables include ampacities for "rubber or thermoplastic insulated" single core cables installed in concrete encased duct banks for groups of 3, 6 or 9 cables, and for 1, 8 and 15 kv insulation levels. At first glance, the 1 kv insulation ampacities appear applicable to dc traction applications. However, it should be recognized that these ampacity values only apply to the specific installations illustrated in Figures 1 through 4 on page VIII of the standard, and for cables carrying three-phase ac current (hence the integer multiples of three cables shown in all the Figures). These installations utilize 5" fiber conduits at 7.5 inch spacings on center; the tops of the duct banks are located 30 inches below grade. The 9-way duct bank shown in Figure 4 is not a practical design. In addition, fiber duct is no longer used Page 3 of 8

5 Rating Considerations for for new construction, and it has a different thermal resistivity than the various thicknesses of the PVC, polyethylene, and fiberglass reinforced epoxy conduits used in modern practice. IEEE Standard IEEE Standard 835 is an enormous volume (over 3000 pages) of cable ampacity tables and supporting information, including equations, constants, and example ampacity calculations. It is essentially an update of the ICEA Black Book whose primary objective was to account for advances in medium and high voltage cable construction since the Black Book was published in Like the Black Book, it utilizes the Neher-McGrath Model for calculation of underground cable ampacities. It also uses similar duct bank installation geometries and assumptions. However, all duct bank configurations for cables insulated at 5 kv and below ("Type 1") are for "triplexed" cable circuits (three cables per duct), a practice rarely utilized in dc traction power systems. A triplexed cable circuit produces approximately three times the heat of a single core cable of identical construction and load current; hence, these values are of no practical use for determining accurate ampacities for single core cables. Single core cable ampacities for 3 and 6 cables per duct bank are provided for 5-15 kv class and higher concentric neutral shielded cables for 75% and 100% load factors; these ampacities would be overly conservative (low) for dc cable application. In summary, this standard is not a good resource for dc traction power cable ampacities NFPA 70, National Electrical Code The 2002 edition of the National Electrical Code (NEC) contains cable ampacity tables in Article 310 and in Annex B. The tables in Article 310 applicable to cables in underground ducts (raceways) have been compiled on the basis of ("up to") three current-carrying conductors per raceway. As with triplexed cables, these circuits produce much more heat than single core cable circuits of the same construction, and have correspondingly lower ampacities. In addition, the tables in Article 310 are for single circuits only. Annex B of the 2002 NEC references both the ICEA Black Book and IEEE Standard It provides only one ampacity table, Table B.310.5, for multiple single core cable circuits with one cable per duct. However, the maximum conductor temperature on which these ampacities are based is 75 Degrees C, not the 90 Degrees C typical of dc traction power cable circuits. Summary of Ampacity Table Review Of the three sources reviewed above, the ICEA Black Book is the only collection of ampacity tables that is applicable to the steady state rating of dc traction cables in duct banks with 9 conduits or less. Utilization of ampacities derived from the Black Book may result in overly conservative cable ratings, and therefore unnecessary construction cost. A much better approach for determining dc traction cable ampacity is described below. APPLICATION OF AMPACITY ANALYSIS SOFTWARE CYMCAP for Windows Software for the simulation of both steady state and transient cable performance is commercially available. Application of this software makes the use of ampacity tables unnecessary, since these programs can incorporate specific installation particulars not found in tables. The application of a program available from CYME International, called CYMCAP for Windows, is discussed below. CYMCAP for Windows is the current version of a program originally developed by Ontario Hydro under contract to the Canadian Electrical Association [4]. CYMCAP utilizes the full IEC 287/Neher-McGrath method, plus IEC Standard 853 for cyclic and emergency rating calculations. CYMCAP can calculate the steady state and transient ampacities for duct banks of any practical configuration containing up to 45 cables. Calculations support different cable types and loads in the same duct bank and even ducts of Page 4 of 8

6 Rating Considerations for different sizes in the same duct bank. In addition to powerful analytical capabilities, the program makes use of a graphical interface to construct models of cables and duct banks. models can be constructed "layer by layer" on-screen, with the appropriate thermal resistivity being assigned to each layer. The cable models can then be saved in the "cable library" for future reuse. Ampacities for riser cables in air or in various conduit types and configurations can also be calculated, if desired. For dc traction power cable applications, the program allows the electrical frequency of the load current to be set to zero; this results in the correct treatment of the cable losses that only occur in the presence of alternating current and voltage. Steady State Analysis Application CYMCAP ampacity cases were performed to replicate the modern equivalent of the geometry and conditions defined in the ICEA Black Book Figures 2 and 3. These figures define installations for 3 and 6 cables per duct bank, respectively, and the corresponding ampacity results are listed on Black Book page 180. Modern 2000V EPR insulated, XLP jacketed, single copper core dc traction power cables of typical sizes were modeled, installed in 5 inch type EB35 PVC conduits (conduit intended for concrete encasement). The CYMCAP cable models were constructed from low smoke, zero halogen traction cable catalog data provided by a prominent cable manufacturer. For the sake of brevity, only 100% load factor and typical RHO 90 earth resistivity were evaluated for a 90 Degree C maximum cable operating temperature. The results of these ampacity simulations, and their comparison with the corresponding ampacities found in the Black Book, are shown below in Table 1. Note that the calculated ampacities are all higher than those found in the Black Book; the differences between the calculated and Black Book ampacities increases with cable size. To investigate the impact of conduit thickness alone on ampacity, the same cases were recalculated using the thicker 5" schedule 40 PVC conduit in place of the original 5" type EB conduit. The results of these simulations are shown in the second column from the right in Table 1; note that ampacity is reduced about 2-3% by the thicker conduit. To investigate the impact of conduit diameter alone on ampacity, the same cases were recalculated using 4" Type EB35 PVC conduit in place of the original 5" type EB conduit. The results of these simulations are shown in the far right column of Table 1; note that ampacity is almost imperceptibly reduced by use of the smaller conduit. Table 1 Steady State Ampacities for ICEA Black Book Duct Bank Configurations # of CYMCAP ICEA % w/5" s in ICEA Ampacity Ampacity CYMCAP Sch. 40 Duct Bank Fig. # () () to ICEA PVC (A) w/4" EB35 () kcmil % % ,118 1, % 1,087 1, ,315 1, % 1,278 1, % % % , % 1,037 1,051 Page 5 of 8

7 Transient Analysis Application Rating Considerations for It is commonly understood in power cable engineering that large diameter cables require a considerable length of time to achieve their continuous rated temperature when carrying rated load current, particularly large cables installed underground. Very little data providing more than very rough approximations for these time intervals is available in the industry literature. The solution of this problem involves a transient thermal analysis, which becomes rather complicated computationally when multiple cables are involved. CYMCAP has the capability to calculate the time intervals required for multiple cables in a duct bank to be heated from a steady state initial temperature to a final temperature. For illustrative purposes, CYMCAP was used to calculate these time intervals for the ICEA Black Book Figure 2 duct bank (2 x 2). In Tables 2 and 3 below, the time intervals required to heat all three cables in the duct bank from an initial to a final 90 Degrees C are shown for two arbitrary values of cable current. These load current values are 125% and 150% of the cable steady state ampacity as previously calculated by CYMCAP in Table 1 above (100% load factor assumed). Table 2 provides results for a condition that could represent the beginning of a morning rush hour after a period of no revenue service (no load or light load), which is a 25 Degree C initial operating temperature. As shown in Table 2, when 125% of rated load current is applied to all cables, it takes between 30.7 to 49.1 hours for the hottest cable to reach its rated 90 Degree C temperature, depending on cable size. If 150% current is applied to all the cables, the heating intervals will range from 2.7 to 5.5 hours. Table 2 Transient Thermal Analysis for ICEA Figure 2 Duct Bank Times to Reach 90 C from a 25 C Initial Temperature Initial Temp. Initial 125% Load Hours to Reach 90 Load 150% Load Hours to Reach 90 Load kcmil C , C 254 1, , C 320 1, , C 377 1, , Table 3 provides results for a 50 Degree C initial cable temperature condition that could represent the beginning of an evening rush hour. As shown, when 125% of rated load current is applied to all cables, it takes between 6.6 and 13.8 hours for the hottest cables to reach their rated 90 Degree C temperatures, depending on cable size. With 150% load current applied, these intervals range from 1.2 to 2.4 hours. Table 3 Transient Thermal Analysis for ICEA Figure 2 Duct Bank Times to Reach 90 C from a 50 C Initial Temperature Initial Temp. Initial 125% Load Hours to Reach 90 Load 150% Load Hours to Reach 90 Load kcmil C , C 605 1, , C 762 1, , C 899 1, , Page 6 of 8

8 Transient Analysis Summary Rating Considerations for The results provided in Tables 2 and 3 illustrate the enormous "thermal inertia" that is characteristic of underground dc traction power cable systems. For these examples, even with a 125% load applied to cables already operating at 50 Degrees C, the cables will never reach their 90 Degree C rated temperature during a rush hour or peak load period lasting from 2 to 3 hours. The cables will heat up even less for a period of catch-up service with the same load currents. These results have significant implications for costeffective dc cable system design: using 100% daily load factor as a design criteria is an overly cautious approach for a typical transit system with morning and evening rush hour peak load periods. These results also help to explain why some older North American transit systems have for many years been successfully using cable sizes and quantities that would be insufficient according to the ICEA Black Book at 100% and even 75% load factors. As noted above, the load factor is a traditional approximation that attempts to quantify the impact of load variation on cable ampacity. With the availability of programs such as CYMCAP, this approximation is no longer necessary to take advantage of the thermal inertia inherent in underground traction cable systems. CYMCAP provides the ability to apply a user-defined 24 hour load curve to each cable installed in a duct bank, and to determine the resulting cable temperatures. The load curve describes, hour-by-hour, how the cable load varies with time of day (rush hour service, "shoulder" periods, non-revenue service, etc). This means the system designer now has the capability to: 1. Select a worst-case design criteria condition such as a feeder breaker or rectifier outage; 2. Simulate the resulting expected rms cable loads using a load flow program; and 3. Verify that acceptable cable temperatures will not be exceeded at any time of day by inserting simulated cable loads into a program such as CYMCAP and performing a transient thermal analysis using applicable daily load curves. This approach is far superior to using ampacity tables, or even steady state cable ampacity simulations with load factors. The results discussed above also indicate that there may be no "continuous load" in a dc traction power system powering rolling stock similar to that defined in the National Electrical Code. Because of the variability of the load and the predictable daily load shape, it can be argued that cable ampacities for large traction cables should be at least somewhat time-dependent. emergency ratings are already recognized as being time-dependent by North American standards. Load current rating classifications for traction rectification equipment are also time-dependent (extra-heavy traction, traction, etc.). According to this argument, cables in duct banks could be assigned short-term and long term ampacities, with the shortterm ratings being applicable to emergency "peak load" conditions such as catch-up service or equipment outages. This could eliminate the well-intentioned practice of oversizing cable circuits for the sole purpose of sustaining short term overloads that, in actuality, would never heat the cables to 90 Degrees C. SUMMARY This paper has provided a brief review of the challenges facing the engineer designing underground cable installations for dc traction power systems. It is apparent that ampacity tables are useful only on a very preliminary basis for this purpose. A cost-effective cable system design that will demonstrably and reliably sustain expected future load levels and equipment outages should be based on a transient cable ampacity analysis. This analysis will take into account the inherent thermal inertia (time lag) of the cable system in response to the daily load cycle of the specific system being designed. Sophisticated and proven software is commercially available to assist the engineer with this effort. Page 7 of 8

9 END NOTES Rating Considerations for 1. George C. Anders, Rating of Electric Power s, IEEE Press, Piscataway, NJ, 1997, Chapter J. H. Neher and M. H. McGrath, The Calculation of the Temperature Rise and Load Capability of Systems, AIEE Transactions, Vol. 76, Part 3, October 1957, pp "Westinghouse Distribution Systems Electric Utility Engineering Reference Book, Westinghouse Electric Corporation, East Pittsburgh, PA, 1965, pages G. J. Anders et al, "FIECAG - A User Oriented Utility Program for Ampacity Calculation of Power s on a Desktop Computer, IEEE Transactions on Power Delivery, Vol. PWRD-2, No. 2, April Page 8 of 8