Prospectus

Return to Table of Contents Volunteer to help write Position Paper

PPP-5 Suitability of Dynamic Models for Relief Valve Loading Expected Benefit toMembers: Engineers will be able to size column relief valves and associate header/flare systems with reduced safety margins, resulting in lower costs. Traditional Research Idea

Present Situation and Proposed Research: The major process simulation vendors should have dynamic model simulators operational during the prospectus period. The dynamic simulation vendors along with the operating companies will “tune” and adjust the simulations basic data during normal operation. However, there will be almost no opportunity to verify that dynamic models accurately predict relief valve loading during a severe unsteady state upset that causes a relief valve discharge. The proposed research involves inviting the major dynamic simulation vendors to participate. Each would generate a model of the FRI test facilities and predict the relief valve loading for common upsets, such as

loss of cooling water or reflux pump. FRI would operate one column (possibly the high pressure) with a temporary relief valve (set to relieve before any existing relief valves) discharging into the other column. The other column’s condenser would be operational. FRI would purposely cause an upset by stopping cooling water or reflux pump. The resulting pressure rise and relief rate would be recorded. FRI members can see how the various dynamic models perform in predicting the relief rate caused by the upset. The dynamic simulation vendors would be free to use the data to improve their simulator (if necessary).

Proposed Internals and Test System: One of FRI’s sieve trays (or valve trays) Estimated Unit Time: Two weeks, one week for steady state tuning of the models followed by one week of upsets for data collection. Estimated Additional Costs (Beyond Unit Time): Unknown. Background and Discussion: Current procedures for column relief valve sizing generally use steady state or over simplified models to approximate the largest expected relief that occurs in an unsteady state system. This is especially true for columns with pump around heat removal systems. Thus, the resulting relief valve and any associated header/flare system are generally oversized. Some relief valve loadings have been calculated by dynamic models. These have resulted in lower than normal relief loadings. However, the validity of the lower relief valve loadings cannot be easily ascertained.

PROSPECTUS

Page | 79