API MPMS 23.2:2020 pdf free download.Manual of Petroleum Measurement Standards Chapter 23.2 Reconciliation of Liquid Tank Car(s) Quantities
Setting fixed limits for L/G, without regard to actual data, may provide performance guides that may be required for contractual reasons. Whenever possible, it is more practical to set limits based on historical data. A measurement system tends to operate at a level of performance that is dictated by physical configuration, equipment, procedures, maintenance practices, environmental conditions, and employee training. All of these factors combine to produce a natural randomness and, sometimes, a natural bias in a system. For systems that have other constraints, such as loss allowance, it may be desirable to include a second set of limits set at the value of the loss allowance.
This would indicate how the system is performing with respect to the loss allowance and if the assigned loss allowance is realistic. It is good practice to determine whether or not a system is stable and in control. A system is generally considered to be in control if the data are all within control limits that have been established from the data. Data points outside the control range indicate poor control. A system is said to be stable if the data exhibit only random fluctuations around the mean without trends.
When physical or operational changes are made to a system, the L/G pattern for the system will often change. When this happens, the prior 2-year history may not be suitable for setting the control limits. In such cases, a moving range chart may be used until sufficient history is developed to define the system’s new pattern. In a moving range chart, the mean and standard deviation are recalculated each time new data are available using all data since the change. The resulting mean and control limit lines on the control chart may exhibit an immediate step change to a new level of control or may change gradually for some period of time until the system stabilizes at a new level of control. As an example, Figure 4 shows three distinct patterns that may be found on control charts.
The points 1 through 7 exhibit random fluctuations around the mean and are well within the control limits. This portion of the data is stable and in control. The points 7 through 12 are within the control limits and appear to fluctuate randomly but are all above the mean. This is a state of stability but not in control because the data do not hover around the mean. In fact, it would appear that the system has attained a new state of control that is centered around a higher mean value. The points 11 through 16 are neither stable nor in control because they are in a definite downward trend. The data do not center around a mean and appear to be headed off the chart.
As a rule, five consecutive points above or below the mean indicate a loss of control or a change to a new level of performance. Five consecutive points trending in one direction (up or down) indicate a loss of control. For some systems, even fewer points in a row may be a significant warning. Examples might be leaking tanks (in which case the losses are real) or meters that are wearing badly and are not being proved often enough (which are book losses). An upward trend is no better than a downward trend.
Either condition is out of control. A system gain can be just as bad as a system loss. Losses and gains occur because of some deficiency in measurement. If the data tend to swing back and forth as shown in Figure 5, the system is cyclic. If the cause of the cycles could be eliminated, the system should be able to achieve a state of better control with narrower control limits.API MPMS 23.2 pdf download.API MPMS 23.2:2020 pdf free download
