To install **StudyMoose App** tap and then “Add to Home Screen”

Essay,
Pages 9 (2243 words)

Views

5

Save to my list

Remove from my list

Characteristically the Petrochemical industry needs machines and equipment to work continuously, and any unscheduled downtimes or interruptions during production time, create massive losses in revenue and hamper the quality of the final product. It would therefore be advantageous to use MFOP as reliability metric and give machines or equipment a set MFOP, within a predefined confidence interval. This would potentially allow for better production scheduling, better downtime planning, could reduce the logistic footprint and yield a steadier throughput within a plant or system.

This methodology is then applied in an industry case study, here the research can be validated and the null hypothesis can be either rejected or accepted.

In the sections that follow the study is performed on data collected from Petrochemicals. A brief outline of the specific problem is given first, thereafter more is said about the case study and the system analyzed is formally introduced. Practicalities of data gathering are then discussed, with data requirements, collection and classification playing a prominent role.

Now that the case study has been defined the analysis can be conducted. Three identical systems were analyzed, general and coherent analysis steps were followed throughout.

The global problem, as stated previously, is that current reliability metrics, Mean Time Between Failure (MTBF) that is widely used, do not provide a unambiguous and untainted view of the performance and operation of the equipment analyzed. MTBF remains a simple mean and has created a widely accepted view that random failures are completely unavoidable. In this specific study it is to be researched if there is an application within Petrochemicals for the proposed solution to the shortcomings of MTBF, Maintenance Free Operating Period (MFOP) originating from the aviation industry.

Having found a research partner, Petrochemicals, a system or equipment needed to be found so that the hypothesis could be tested. After a visit on-site and discussions with various key players from the plant, a system was established. The equipment that was chosen to be studied was a grouping of cone crushers used in ethylene operations. The group of Pumps consisted of three individual and identical pumps, Turbine and Compressor. These pumps were vital to the overall process. Due to smaller sized pumps, Turbine and Compressor in the process, a failure at this station in the process would have had far reaching consequences in proceeding stations. Another factor that was taken into consideration was that this grouping of pumps, Turbine and Compressor had seen some unreliability in the past, a better picture of they was therefore desired.

The aim of this Case Study is to investigate the idea of using the aviation derived concept of MFOP within a Petrochemicals environment. Here a specific item of the total Petrochemicals system is chosen and its failures are modeled and the MFOP concept then applied to the chosen item, in this case a grouping of pumps. The failure distribution of the pumps, Turbine and Compressor is found and thereafter the relevant MFOP statistical formulas are applied on the distribution, in order to find the maintenance free time of operation, specific for that item. It can then be seen if this period of time is feasible and could be applied to the item. The modeling of the pumps, Turbine and Compressor failure distributions is an additional sub-aim of the study additionally. Ultimately, a better understanding of applying the MFOP concept should be gained, thereby testing the school of thought of MFOP against that of the traditionally used reliability metric MTBF.

After detailed discussions with the concentrator plant manager on the system and its subcomponents, a decision was made on which equipment to use for the analysis. This decision was based on the plant manager’s in-depth knowledge of the complete system and the complexities thereof. Also taken into account, were conditions such as equipment criticality to continuous uninterrupted operations, based on non-operational output loss. The chosen equipment was an arrangement of three pumps 201 A-B-C. These pumps were pivotal to the smooth operation of the complete system and the equipment downstream from them. They were, however, to different degrees, susceptible to breakdowns and therefore unplanned maintenance activities had to be conducted, thereby hampering the continuous operation of the system. Due to their susceptibility and varying reliability, this system was chosen for the study. Figure 3 shows the methodology that was already put. Ethan enters this unit as a feed of this unit from the adjacent petrochemical plant. This feed comes with the ethanol from the unit itself into the cracking in furnaces. The furnace output is immediately cooled and into the compressor section.

The moisture content of the exhaust gas is taken from the compressor in moisture absorbers and then enters the cooling section. After cooling, cracking gas enters the distillation towers to separate the various carbon slices. Finally, the C2 Cut tower enters the hydrogenation reactor and after converting acetylene to ethylene to the Ethylene separating tower, ethylene is distilled from ethanol, and pure ethylene is sent from the top of the tower for use downstream of the polyethylene or storage tank for export. The data that was used in this study were failure and maintenance data of the system of pumps 201. This data come from secondary sources, this being from petrochemical data recorder system, which records vast amounts of data and variables. From this system maintenance data on the pumps were pulled and stored in Microsoft Excel. Data was obtained from 2016 to 2017, at a resolution or in increments of 10 minutes over the entire period; a higher resolution would have become impractical within the Microsoft Excel environment, due to the large amounts of data points. Data could not be obtained earlier than the 1st of October 2015, due to no data on failures being available before that point. Table 3 shows the number of events found for each pump from the stated period. The first calculation performed on the data was the Laplace trend test. The result of the Laplace test for the found data was UL-P-201-A = −2. 5648, this was clearly in the reliability improvement area of the test and therefore displayed a trend, therefore making the Lewis–Robinson trend test superfluous. The results are outlined in Table 4. The Power Law NHPP was used in this analysis. The data set for Pump 201-A can be analyzed using repairable systems theory. Here a power law NHPP was used to model the system in order to find expected failure times. The power law’s parameters, λ and δ, were found through the least-squares method, the parameters were found by using the solver function in Microsoft Excel. From Figure 8 it can be seen that Pump 201-A did not yield particularly high MFOPS probabilities for long MFOP. The comparison of the historic MTBF, to the equivalent MFOP of the same length gave the following results that were found through the application of failure statistics and are not just a mean. At Pump 201-A’s MTBF of approximately 57. 12 hours, the probability of achieving this, or the MFOPS, was about 53 %, without requiring any corrective maintenance actions. A further MFOP length that could be taken, as it is easy to comprehend, is the length of a full day or 24 hours. Inspecting Figure 8, it is found that Pump 201-A had a MFOPS of 70 % at an MFOP of 24 hours.

MFOP Calculations for Hypothetical Pump 201-A In order to better perceive the MFOP principle, a hypothetical Pump 201-A was modeled. This crusher used the same data set found for pump 201-A but removed the top two failures, from the data set. Even though this is a hypothetical system, it does not seem unrealistic in the targets it achieves. In the case of pump 201-A, the top two failures that were removed were the categories of the lube system and feedback faults, these made up approximately 70 % of the failures of Pump 201-A. Now that a new data set was formed, the analysis began again with the application of trend tests to the failure data set. Using the Laplace trend test, it was established that the hypothetical pump 201-A laid in the grey area of the test. Therefore, no conclusive statement can be made as to whether their is a trend present or not. The Lewis–Robinson trend test was then applied to the data set, here, again no conclusive result was found, the data was still in the grey area. At this point an assumption needs to be made, plotting the data, it can be seen that the system looks like a repairable system with a distinct reliability improvement; it was therefore decided to model the system accordingly. The power law NHPP was therefore used to model the new data set, with the power law parameters, λ and δ, being found numerically, using the least squares method. The results for the parameters are shown in Table 7. After the parameters for the power law NHPP were known, the MFOP analysis could begin. Plotted in Figure 7, is a comparative plot of the MFOP length and equivalent MFOPS for both the current Pump 201-A and the hypothetical Pump 201-A system. Analyzing the found shape parameter, β, of PUMP 201-B, which is also known as the Weibull slope, as the value of β, is equal to the slope of the probability density function shown in Figure 9. Even though PUMP 201-B is not a non- repairable system, it behaves like one, with the current β > 1; PUMP 201-B displayed a probability of failure that decreases with time. Analysis determined the scale parameter, η, which has the effect of stretching out the probability density function, or the same effect as a change in the abscissa scale. The peak value of the probability density function curve could decrease, as the area under the probability density function remained a constant one. An increase in η, while keeping β constant, stretches the curve out towards the right and its height decreases. A decrease in η, while keeping β constant, pushes the distribution to the left and its height increases. Now that parameter estimators are known and systems reliability is modeled, including graphically, the MFOP calculation can begin. Comparing the current MFOP of PUMP 201-B and the hypothetical one in Figure 10, it is immediately noticeable that the hypothetical system provided a substantially improved MFOP, at a far higher probability of success. Studying the 64. 98 hour MTBF value that was established previously, it now becomes apparent that the new system has a far higher chance of achieving this period, up from 50 % to 92 %.

Summary of Results ofPUMP 201-BP-201-A and PUMP 201-B have very similar MTBF values, 57. 12 hours and 64. 98 hours respectively, but behave very differently and need therefore to be modeled in a different way. The greatest cause of failure for PUMP 201-B, as for all pumps, is the rotor system, accounting for nearly 50 % of the total failures within the analysis period. Owing to the results of the Laplace and Lewis-Robinson trend tests, the pump was modeled using the Weibull distribution with parameters found, as seen in Tables 10 &12. The MFOP calculations performed on pump two found that the crusher at an MFOP length equal to that of its found MTBF had a MFOPS of approximately 50 %. Therefore giving the pump a 50 % chance of achieving its found MTBF. In order to show another perspective of the PUMP 201-B system, a realistic hypothetical case was constructed. The top two failures found in the Pareto analysis were removed from the data set and a new failure data set was compiled. The As with PUMP 201-B, P 201-C does not provide a particularly high MFOP at a high probability of achievement. However, immediately noticeable, is that P 201-C is more reliable than PUMP 201-B. Comparing P 201-C’s previous MTBF, 48 hours, with the found MFOP it was seen that P 201-C actually only had an approximately 50 % chance of completing this period without requiring any corrective maintenance actions. As with PUMP 201-B, looking at a 24 hour period, is relatively easy to comprehend, Figure 13 shows that the pump had a 50 % chance of completing a full day without requiring corrective maintenance actions.

MFOP Calculations for hypothetical PUMP 201-CA hypothetical P 201-Cwas modeled, again, in order to better demonstrate the MFOP concept. Even though this system was hypothetical it was not unrealistic. With reference to the Pareto chart shown in Figure 10, the hypothetical case removed the top two failure cases from the data set. In the case of P 201-C, these were the categories of lube system failures and other/unplanned maintenance. The latter failure was classified by failures such as sequence stops or data points that are simply named “unplanned maintenance”. These two failure categories made up over 70% of failures for P 201-C. Once a new data set had been formed, the analysis started again with a trend test. The Laplace trend test established that there was no underlying trend present in the new data set and therefore the hypothetical pump could again be modeled using the Weibull analysis. Weibull parameters were then found by using the Maximizing the Likelihood Method, parameters β and η, shown in Table 16. As the Weibull parameters, β and η, of the hypothetical system are now known, an MFOP calculation could be performed. The results were plotted together with the current MFOP performance, shown in Figure 14. By comparing the two systems, a vast difference can be seen in the probability of achievement of MFOP length. The supposed system was immensely more reliable than the current system. Taking the previously established current systems MTBF of 48 hours, P 201-Chad a 37 % chance of completing this period. The proposed system would now have a 57% chance of making the same period, a sizeable improvement.

Live chat
with support 24/7

👋 Hi! I’m your smart assistant Amy!

Don’t know where to start? Type your requirements and I’ll connect you to an academic expert within 3 minutes.

get help with your assignment