ML19296A888

Comments on Nutech,Inc Smoke Detector Siting Pilot Test.Test Method Was Applied W/O Appropriate Engineering to Develop Technique

ML19296A888
Person / Time
Site:	Yankee Rowe
Issue date:	11/09/1979
From:	Fung F AEROPHYSICS CO.
To:	Ferguson R Office of Nuclear Reactor Regulation
Shared Package
ML19296A886	List: ML19296A884 ML19296A888
References
NUDOCS 8002190228
Download: ML19296A888 (3)

Text

g Enclosure No. 2 l

M_

AEROPHYSICS COMPANY

. g ICS 3500 Connecticut Avenue, N.W. / Washington, D.C. 20008 / Telephone (202) 244 7502 9 November 1979 Mr. Bob Ferguson NUCLEAR REGULATORY COMMISSION 7920 Norfolk Avenue Bethesda Maryland 2001'.

Dear Mr. Ferguson:

The Nutech smoke detector siting pilot test only proved the feasibility of applying the well publicized SF6 tracer gas technique to the smoke detector siting problem. But as to a test method to yield smoke detector siting recommendations it falls short of its objective badly. Haste to apply the pilot test as a model for smoke detector siting without appropriate engineer-ing to develop the technique as a test method will only lead to inconclusive results and thus guarranteed to waste a great deal of money and effort.

As you know the SF6 technique in the present form was applied successfully to the study of smoke movement and control in high rise buildings.

In thi case because of the strong preferred vertical smoke =ovement the problem is essentially one dimensional. Each vertical shaf: has its maximum refer-ence concentration at the bottom of the shaft. Thus each shaft can be studied on its own.

This effectively removes the complication of the vari-ation of fire simulation location and minimizes horizontal smoke movement effect as secondary.

In snoke detector siting we must face the complexity of the full three dimensional smoke movement problem. This introduces among others the need to visualize horizontal flow movement patteras and their interactions, proper scaling of simulated fire size in relation to ceiling height and pressure and HVAC flow parameters for fire plume dynamic simulation, and the need to remove anomolies introduced by variation of similar fire lo-cations.

To sum up a properly engineered smoke detection siting test method must be:

a) Insensitive to fire simulation locations, b) Universally valid with reference concentration for data comparison and interpretation, c) Repeatible and provide deterministic flow pattern visualization without the need for multiple tests as in the pilot test, d) Consistent by providing scaling for simulated fire size in relation to 8002190)

Enclosure No. 2 Mr. Bob Ferguson Page Two NUCLEAR REGULATORY COMMISSION 9 Nove=ber 1979 smoke sink strengths and ceiling heights for dynamic similarity and consistent stratification, e) Economically feasible in terms of test set up and data collection.

(Proper engineering based on items (a) to (d) can drastically reduce the number of data points, and tests needed compared to the pilot test.)

The pilot test did not address any of the above criteria and in no way can be qualified as a test method.

I am enclosing a statement of my evaluation of the pilot test as a test method.

I will be glad to discuss my suggested satisfactory solutions to every criteria listed above.

I believe with little added engineering effort we can have a good test method. A qualified engineer with knowledg in fluid dynamics and experience in smoke simulation and test method devel-opment with my input can without question develop a catisfactory test method. To satisfy the added criteria listed in this letter beyond the art of the pilot test one should not need more than 6 man-months engineer-ing development. Time wise this can be shortened to 3 months if two com-petent engineers work together as a team.

Sincerely, AER0 PHYSICS COMPANY ku r < '.

% 1

~-

Francis C. W. Fung 3

Manager, Energy Division I

FCWF:sb

+

Enclosure No. 2 7

KS Comments to Smoke Detector Siting Pilot Test by Francis C. W.

Fung

1) SF6 smoke simulation testing method was original:y developed and perfected for vertical smoke movement and control in high rise buildings. This is well documented in leiterature. Concentration measurements of infiltration are normalized with respect to the maximum at the bottom of the s' aft.

There is a preferred known direction of smoke movement.

Each shaft can be studied on its own, insensitive to fire location, or multiple paths of smoke.

Problem is reduced to simple one dimensional, in addition fire size is not critical hocause each shaft has it's own maximum reference concentration at the bottom of the shaft.

2) Application to smoke detector siting in both horizontal and vertical smoke movement 14 new, thus need close examination.

In a complete 3 dimensional situation such as for smoke detector siting we face the difficulty of large numbers of important unknowns. These unknowns such as 3 dimensional complex smoke flow paths, flow path interactions, strength and location of smoke sinks depend on a large number of data needed to be collected for proper interpretation of results. The present sampling technique means large man-power requirement, thus is not economically feasible as a test method.

(The pilot test took a team of over 10 engineers to provide the data.)

3) The pilot test raised more questions than provided answers. As a test the results are not uniformly valid. Poor resolution, inconsistencies, most of all lack cf a reference concentration for data are caused by lack of under-standing of low patterns and their interactions, not knowing the relative strength of smoke sinks (exhausts), and inability to determine fire plume size to extablish consistent stratification. Finally the test results yielded by the pilot test set up is far too sensitive to fire simulation locations.

4) From lessons learned from pilot test a properly engineered smoke detection test method must be:

a) Insensitive to fire simulation locations, b) Universally valid with reference concentration for data comparison and interpretation, c) Repeatible and provide deterministic flow pattern visualization without the need for multiple tests as in the pilot test, d) Consistent by providing scaling for simulated fire size in relation to smoke sink strengths and ceiling heights for dynamic similarity and consistant stratification, e) Economically feasible in terms of test set up and data cellection.

(P oper engineering based on items (a) to (d) can drastically reduce t a numbe r o f da *a poin ta and testa needed compared to the pilot test.)