The electric power system has been getting smarter for decades, as new technologies allow better analysis and greater control. But most utilities have implemented these technologies in a piecemeal...
Assessing Construction Compliance
Gas utilities can make better use of their inspection budgets.
and resources, and misreads the risk.
Achieving 95% Confidence
Determining the importance of each of the 103 compliance factors involves polling a group of company people at several levels and from different parts of the organization, with varying degrees of specific knowledge about construction, on which elements the group believes to be most critical if failure occurs. This yields a set of weighted criteria that provides the basis for a single risk-adjusted compliance score for each installation inspected. This has been found to be a surprisingly accurate predictive tool.
In the past, field auditors dropped in on an installation and observed a small portion of whatever was underway at the time. The new program is different in several important ways. Namely:
• Auditors observe installations from beginning to end;
• Sites are assigned auditors through a carefully randomized process;
• Should compliance problems emerge, auditors are instructed to record the as-found situation, require the crew to correct the problems, and make certain the installation is in compliance;
• Field audits are unannounced, addressing the principal elements associated with risk to people and property introduced by non-compliant construction practices.
Besides inspecting installations from start to finish, and assigning auditors to installations at random, the question arises: “How many installations need to be inspected to produce compliance measures that are highly confident?” Cost must be balanced with the degree of risk the company is willing to undertake. The risk is measured by the compliance level. The cost is determined by the required number of samples, which in turn are determined by:
• Size of the installation/construction program: In this example the client had 25,000 new installations annually;
• Confidence Level: Desired level of statistical significance (usually a 95-percent confidence level);
• Required precision: How close should the sample predict the true (but unknown) risk and compliance value for the system; and
• Inherent variation: The true value of risk or compliance for the system. This isn’t known in advance, but requires an initial experienced-based estimate (see Figure 2) .
For the client with 25,000 new installations to be inspected annually, focused inspections on about 505 sample installations produces compliance measurements for the entire system at the desired 95-percent confidence level (see Figure 3) .
To get a 100-percent confidence level—and 100-percent precision on the actual level of risk and compliance— it would be necessary to inspect all installations, a sample size of 25,000. Choosing a 95-percent confidence level and a 2-percent precision level allows very high confidence that the estimate truly reflects the actual level of risk, but at a substantially diminished cost.
The significance of this statistically valid confidence level is that it’s widely accepted by industry regulators, federal and state courts, as well as the military, because it is recognized as being as good as reasonable people acting prudently can achieve— i.e., the common-law “prudent-man rule.” The Nuclear Regulatory Commission accepts this standard on such things as stainless-steel welds in nuclear power plants, and the armed forces accept this on deliverables such as bullets. Many courts have upheld the prudent-man standard