Reliability Evaluation of Rotary Compressor by Two-Parameter Step Stress Life Testing

One of major failure causes of rotary compressor is lubrication-related. Under the premise that bearings involve no major faults either in the design and material selection, their reliability is sensitive to the amount of system refrigerant charge and the amount of oil. These two factors, oil and refrigerant in the oil sump of rotary compressor are of great interest for lubrication failure modes and their causes. This paper introduces a two-parameter step stress life testing methodology for evaluation of lubrication-related reliability. The methodology uses the amount of oil in the compressor oil sump and refrigerant in the refrigeration system as the two control parameters. Refrigeration oil, as one parameter, is reduced in increment, while refrigerant charge is increased in step-by-step increment In such a step-wise increase in stress, lubrication failure is bound to occur. Resultant failure modes and the stree level, that is determined by the combination of the two control parameters provide design, reliability and application engineers with valuable information. Applications of the step stress methodology include evaluation of bearing design, material selection, and system design, for examples. TEST STRESS and ACCELERATED LIFE TESTING It is a common practice in the field of compressor reliability engineering to use some forms of accelerated life testingY1 One common way of accelerated life testing is testing compressor at a fixed stress level during the entire testing period. The definition of reliability relates to the operating condition that, in turn, determines stress level imposed on compressor. The stress used in this context is not the conventional definition of stress (force per unit area), defined in mechanics of materials, as the internal reaction of a body to the application of external loadsP1 It rather means that the change in test condition or test variables (such as amount of oil or refrigerant) imposes change in stress level which a compressor is subjected to. LUBRICATION-RELATED FAILURE MODES and CAUSES In rotary compressor, lubrication-related failures modes are shown basically in journal bearings and roller OD surface. Rotor drag against stator ID, scored cylinder ID, scored or worn shaft journal are also observed as primary or secondary results. In most cases, lubrication-related failure modes are fairly self-evident. The failure fundamentally comes from insufficient or breakdown of oil film thickness. The problem of oil film can come from refrigerant content in oil, mechanical bearing load and oil temperature. All these are related to the film thickness. Causes for the failure modes are lack of oil, relative refrigerant over-charge leading to low oil viscosity and insufficient oil film thickness. Fifteenth International Compressor Engineering Conference at Purdue University, West Lafayette, IN, USAJuly 25-28, 2000 161 THE CONCEPT OF STEP-STRESS TESTING The concept of step-stress testing is to subject a component or product to increasing stress level in each subsequent step of the testing. The test stress may be in the form of pressure, temperature, quantity, mass, etc. Each increment can be determined by percentage of the standard or a certain absolute amount. An example of such a testing is a hydrostatic strength test where a vessel is exposed to a pressure and held for a period then moved to the next higher pressure level for a pre-determined duration. If the vessel holds up successfully then the pressure level is increased to the next level and the test continues on until the vessel structurally fails. This simple example is one-parameter step-stress testing. On the other hand, twoparameter step-stress testing would be testing compressor subjecting it to two completely independent stress parameters such as amount of oil and refrigerant. In each subsequent step, oil amount is decreased while amount of refrigerant is increased. In this way, the stress applied to compressor is increased stepwise. Selection of these two parameters will result most likely in a lubrication-related failure mode. SELECTING TWO FUNDAMENTAL PARAMETERS The design of journal bearing is based on the premise that the lubricant will form a fluid film separating two surfaces (journal OD and bearing ID) which slide relative to each other. Though bearings are designed to maintain the fluid film under most of the encountered operating conditions, but when the complete oil film cannot be maintained a metal-to-metal contact occurs and bearing failure followsF1 To have a thicker oil film, viscosity of the lubricant must be increased, but refrigerant affects viscosity and temperature of the lubricant depending on the extent of refrigerant dissolved in the oil. Therefore, the ratio and composition of the two, oil and refrigerant are of great interest to bearing life. There are many parameters which determine "stress" on bearing lubrication including type of lubricant, operating lubricant temperature, bearing design, bearing clearance, material, surface finish, etc. However, in this paper, focus is on the two key operating parameters. They are the amount of oil in the oil sump and amount of refrigerant charged in the refrigeration system. These two parameters are key factors given other parameters are fixed. And the two parameters define the quality of oil. Reasons for selecting the parameters are 1) they are easily deviate from the standard specification and 2) they are also two major contributors to defining the quality of oil. Amount of Oil The total amount of oil charged into compressor at the compressor factory does not stays in the compressor oil sump in an operating unit. The "wetting" of interior surface of tubing, other system components and inside compressor comes from the original total charge amount. Amount of oil in compressor oil sump plays vital roles. First it determines oil level that affects oil circulation rate. It also determines refrigerant/oil ratio and determines temperature level. If the oil level drops too low oil may no longer be delivered to bearings in sufficient quantity. Inspection of many field-failed compressors suggests that many compressors lose oil from a variety of reasons. Furthermore, losing oil out of compressor during a normal course of compressor start-up, defrost and start-stop cycling is not unusual. Amount of Refrigerant Unit manufacturers specifies a standard refrigerant charge that gives the system the desired unit performance and compressor protection. However, there are many chances where the standard charge is altered at field site. Servicing the unit, development of leak, overcharging, etc. all add to the complexity. Experiences in the field Fifteenth International Compressor Engineering Conference at Purdue University, West Lafayette, IN, USAJuly 25-28, 2000 162 suggest that distribution of refrigerant in units varies despite a fixed amount is pre-charged. If a unit is field charged the variance is likely more widely distributed. In field units, seeing refrigerant-overcharged units by 25% is not all that uncommon in some parts of the world. This is mainly due to the fact that field charging methods practiced in the field are not very accurate. In actual system, on the other hand, charging beyond 200% of the standard charge is very unlikely. Therefore, testing beyond 200% refrigerant charge may actually lead to unreasonable consequences, especially when the amount of oil stays low. When worse of the two parameters combine the compressor is now subjected to higher stress level than the standard condition. DESIGN OF THE EXPERIMENT A sample test plan may consist of six (6) steps only. For example, the amount of oil, the first parameter, may be set at 125%, 100%, 75%, 50o/o, 25% and 13% of the original oil charge amount. In this case the minimum is set at 13% below which oil level does not reach the oil pick-up tube. Testing a compressor below this minimum level is meaningless: testing with less than 13% oil offers little information. The second parameter, amount of refrigerant, may be set at 75%, 100%, 125%, 150%, 175% and 200% of the standard system charge amount. The stepwise test is then conducted according to the following schedule, running the test for a fixed duration (such as 300 or 500 hours) at each step. Amount Amount of Fraction STEP of Oil Refrigerant of Oil* Test Duration Stress Level 1st 125% 75% 0.53 300 hrs Lowest 2nd 100% 100% 0.41 300 hrs 3rd 75% 125% 0.29 300 hrs 4th 50% 150% 0.12 300 hrs 5th 25% 175% 0.09 300 hrs 6th 13% 200% 0.05 300 hrs Hi est Mass Fraction of Oil/( oil+R22) & Based on Specific Gravity of Alkyl Benzene Oil 0.87 glee Also, Based on Standard Oil Charge of 550cc & 700g R22 Number of Steps There is no fixed number of steps to follow. There can be as many steps as convenient. If the number is one, then the test is at a fixed condition, and this is a common test scheme. Normally, the number is likely to be limited (3 10) for practical reasons. As the number of steps increases there will be more interruptions during the test and the testing takes longer, but test may give you more information. If the number increases more than ten, the frequent interruption during the course of testing may introduce chances of introducing other variables that are unintended but could affect the outcome of the test. In addition, as test duration of each step becomes long the entire test duration becomes unacceptably too long for a timely decision-making. Test Duration The test duration should consider the number of test steps, severity of stresses at each step, total testing time available before the decision, and expected life of the test compressor. If the test compressor survives the intended test without a failure, the test duration may be extended to failure or terminated at a pre-determined time and compare the condition of the tested hardware. If the operating condition at which failure is suspected, that particular step may be followed until failure occurs. In most of cases, test duration of 300~500 hours Fifteenth International Compressor Engineering Conference at Purdue University, West Lafayette, IN, USAJuly 25-28, 2000 163 range is convenient. This particular test duration at each step makes the decision from the test result possible in 2,000~3,000 hours. Ifthe likelihood ofthe combination of refrigerant and oil amount occurs can be estimated, a prefixed test time can be apportioned variably according to the likely probability rather than running the test for the same duration at each step. Load Condition In this particular example the test condition is Max. (High) Load condition with suction flooding. The reason for selecting the condition is oil film thickness is worst within approved application envelope. £31 Furthermore, selection of the two-parameter fits just right to the lubrication-related test purpose. Operating Condition There are several important test data to be collected. Both suction and discharge pressure, suction temperature and oil sump temperature measured at the dead center of bottom shell are minimum data to be collected. From these, percentage of refrigerant in oil and viscosity of oil can be estimated[4]. Then, these data can be used to determine oil film thickness. The test involving the two particular parameters is likely to lead to a lubrication failure. Thus, in this case, test condition should be tailored to fit the situation. Based on an unpublished work[3], the run condition that results in the thinnest oil film is at high load condition at which both suction and discharge pressure is high. On the operating envelope of compressor this point is can readily be located. A LABORATORY TEST SAMPLE For an illustration purpose, an actual test case is shown in Table 1. This test compares bearing materials, A vs. B. Pictorial representation of the twelve-step testing is shown in Fig. 1. In this example, the material B outlasted the material A (Fig. I). From step 4 and beyond, miscibility of R22 approaches to 99.9% in oil, and theoretical dynamic viscosity stays below 0.1 centi-Stokes. The oil sump temperature, represented by the center bottom portion of bottom shell also drops to 58+/-3C from 75+/-5C. It becomes evident that starting the step 4 or beyond stress level to compressor drastically increases as evidenced by oil sump temperature and oil viscosity. Both failures were lubrication-type, represented by distress, either wear or heavy scoring, observed on bearing and shaft. DISCUSSIONS Oil vs. Refrigerant RatioThe mass ratio, oil/(oil+R22) drops quite steeply, 11% per every step up to the 5th step but at much slower pace (0.6%) up to 12th step (Fig. 2). This tapering of · the slope, 4% point drop between step 5th to step 12th, is due to the fixed oil amount. The oil amount lower than certain oil level (below 25 cc) does not provide sensible information. Bottom Shell Temperature Bottom shell temperature is a key indicator for quality of oil in the oil sump thus in estimating oil viscosity and oil film thickness. The rate of the temperature drop is rapid (5~8 C per step) until 4th or 5th step (Fig.3) for the same reason discussed above. According the plot the compressor A was slightly advantageous in oil viscosity though compressor B outlasted compressor A. Percentage of R22 in the Oil Sump The percentage of R22 in the oil sump is calculated from discharge pressure and oil sump temperature using a software available. £41 After 3m step, the percentage approaches practically to 99.9% (Fig. 4). Fifteenth International Compressor Engineering Conference at Purdue University, West Lafayette, IN, USAJuly 25-28, 2000 164 Viscosity of OilSince viscosity depends on discharge pressure and oil temperature that is estimated by bottom shell temperature, the curves (Fig. 5) shows similar but inverse pattern as in Fig. 3. Statistical Analysis of Test Data To have a statistical significance of decision making based on test results one single set of data is not sufficient. Test with more samples are necessary. Thus this type of testing can be used in initial engineering evaluation and quick comparison with limited number of samples. One drawback of conducting this type of testing is that the testing takes time and has numerous interruptions in reducing oil and adding refrigerant. As this type of evaluation is a tedious process requiring substantial laboratory resources, the number of test samples is bound to be limited. Thus, engineering decision-making is likely to be made with a limited number of samples. A simple comparison of time-to-failure data is not appropriate as each subsequent test step beyond the baseline time-to-failure is at higher stress level (or more-accelerated testing). This acceleration fact makes engineering decision possible with limited data available. SUMMARY & CONCLUSIONA step-stress test methodology is introduced using two-parameter scheme. Bothoil and refrigerant amounts are two critical parameters that determine the quality of oil thatfeeds the bearing. An example of experimental design for rotary compressor has beendesigned, tested and results analyzed. Failure at 5th step with one material versus failure at 12thstep with another material is quite significant in that test stress level beyond 5th step increases ineach of the subsequent steps. The testing methodology can widely be used in compressorevaluation. Application of this testing methodology is wide; relative reliability of differentmanufacturers, different processes, material, and designs can be relatively evaluated. Also thistype of test can be used for development and qualification test. REFERENCES1. Ireson, W. Grant, Clyde F. Cooms, Jr, Richard Y. Moss, 2nd ed. 1996. Handbook ofReliability Engineering and Management. New York: McGraw Hill2. Baumeister, Lionel S., 7th ed. 1966. Standard Handbook for Mechanical Engineers,New York: McGraw Hill3. Unpublished Research Report. 2000. Daewoo Carrier Corp.4. Unpublished Software "Nisc2g20," 1999. United Technologies Carrier Corp. Fifteenth International Compressor Engineering Conference atPurdue University, West Lafayette, IN, USAJuly 25-28, 2000165 Table 1 Step-Stress Flooding Test Under High-Load Condition Oii&R22;Average Run ConditionCompr. BearingStep Oil&R22; R22PsPdCompr. Temp.(C)I.D MaterialOilR22 Viscosity Test Duration %%Ratio(a)Kg/aiG KgiCJIG: Psi A Sac. Dist:b. Top BoL(%)** (eSt) (IIrs!Hrs***)ccg I 125 688 75 525 0.53 6.2 24.0 356.1 12.8 70.5 8578 35.6 1.308 300/3002 100 550 100 700 0.41 6.4 26.6 393.0 12.8 71.6 84 75 48.6 0.738 300/300AA 3 75 412 125 875 0.29 6.8 29.0 427.2 127 69.0 87 71 88.1 0.149 300/3004 50 275 150i 1,050 0.19 6.3 27.0 398.7 128 70.4 6860 99.9 <0.1 300/300 5 25 140 175i 1,225 0.09 6.5 27.1 400.1 12.9 70.4 65 58 99.9 <0.1 "24/300I 125 688 75 525 0.53 6.3 27.8 410.1 12.8 69.286 80 43.7 0.870 300/300 2 100550 100 700 0.41 6.5 27.0 398.7 12.8 69.5 86 75 50.4 0.676 300/3003 75 412 125 875 0.29 6.9 28.3 417.2 12.8 68.9 86 69 93.4 0.126 300/300 4 50 275 150i 1,050 0.19 6.9 26.4 390.2 12.8 71.4 66 55 99.9 <0.1 300/3005 25 140 175 : 1,225 0.09 6.8 27.1 400.1 128 70.4 65 56 99.9 <0.1 300/3006 25 140 200 : 1,400 0.08 6.8 27.4 404.4 12.7 70.7 65 56 99.9 <0.1 300/300BB7 25 140 225 : 1,575 O.o? 6.8 27.5 405.8 12.9 71.0 65 57 99.9 <0.1 300/3008 25 140 250 i 1,750 0.07 6.8 28.4 418.6 128 71.1 64 59 99.9 <0.1 300/3009 25 140 275 : 1,925 0.06 6.8 28.6 421.5 129 71.4 64 60 99.9 <0.1 300/30010 25 140 300 : 2,100 0.05 6.8 28.4 418.6 12.8 71.1 64 59 99.9 <0.1 300/300II 25 140 325 : 2,275 0.05 6.8 28.4 418.6 128 71.1 64 59 99.9 <0.1 300/30012 25 140 350 : 2,450 0.05 6.8 28.4 418.6 12.8 71.1 64 59 99.9 <0.1 *52/3008 Notes : [ Flood-back Condition = lligh-l.Aiad( 5511SS/55"F) I => [ Ps/P-.S/28.5 Kg/cm2, Ts=12.8 t I (a) = Oil/ ( Oil + R22 ), Specific Gravity of Oil = 0.871*=Failed 'fime ( Hrs/Hrs Planned), **=Percentage of R22 in Compressor Oil-sump, *** = TestedfPlaoned, <=Max. a. OilJ!!rn (cc)