Page 1 of 5

European Journal of Applied Sciences – Vol. 11, No. 3

Publication Date: June 25, 2023

DOI:10.14738/aivp.113.14608.

Partom, Y. (2023). Diameter and Thickness Effects Scaling Deviations of High Explosives. European Journal of Applied Sciences,

Vol - 11(3). 155-159.

Services for Science and Education – United Kingdom

Diameter and Thickness Effects Scaling Deviations of High

Explosives

Y. Partom

18 HaBanim, Zikhron Ya'akov 3094017, Israel

ABSTRACT

One way to characterize the sensitivity (or the reaction rate) of explosives is

through size-effect tests. For explosive rods they are called diameter effect tests,

and for explosive plates, they are called thickness effect tests. With high reaction

rate explosives, different test configurations usually yield almost the same reaction

rate or detonation velocity. But with low reaction rate explosives, different test

configurations usually yield somewhat different detonation velocities. Following

one of our previous papers [1], we propose here that those different detonation

velocities result from the phenomenon of partially reacted boundary layers, that

form when a detonation wave is grazing along a free boundary. In what follows we

perform computer simulations to show how such a phenomenon comes about.

INTRODUCTION

One way to characterize the sensitivity of explosives (in the sense of reaction rate), is through

size effect tests. Using cylinders (rods) they are called diameter effect tests, and using wide

plates they are called thickness effect tests. From the steady state equations of motion for the

two configurations it can be deduced that these two test configurations scale as:

( ) ( )

( ) ( )

d D 2h D

or

r D h D

=

=

(1)

where d=rod diameter, r=rod radius. h=plate thickness and the scaling hold for the same

steady detonation velocity D,

But tests on various explosives show deviations from the above scaling. Those deviations are

quite small for so called ideal explosives that have very high reaction rates, or very short (of

the order of few nanoseconds) reaction times. But they can be substantial for less ideal or non- ideal explosives. Tests on various explosives show, that sometimes we may get r/h<1, and

other times r/h>1, and that deviations from perfect scaling may be as high as 10%. Searching

the literature, we haven’t seen any explanation of these scaling deviations. Here we attempt

such an explanation. Our explanation is based on previous work of ours [1], where we show

that a boundary layer of partial reaction usually forms when a detonation wave is grazing

along a free boundary.

Page 2 of 5

Services for Science and Education – United Kingdom 156

European Journal of Applied Sciences (EJAS) Vol. 11, Issue 3, June-2023

Denoting the equivalent thickness of such a boundary layer by r for a rod, and by h for a

plate, we have:

( )

( )

( )

( )

r r r 2 h r 1 1

h h h h

so that :

r D 1 for 2 h r

h D

r D 1 for 2 h r

h D

−   − 

= = −

− 

   

   

(2)

On this basis we suggest that scaling deviations of the diameter effect from the thickness effect

are related to the partially reacted boundary layer phenomenon.

SIMULATIONS

To establish the above suggestion, we ran direct numerical simulations of the rod and plate

geometries, using our reactive flow code that we call TDRR (Temperature Dependent Reaction

Rate) [2]. As we’ve shown previously, by using our TDRR model we’re able to reproduce the

partially reacted boundary layer discussed above. It follows, that if TDRR is also able to

reproduce the above-mentioned scaling deviations as well, this would strengthen our

suggestion mentioned above for the origin of those scaling deviations.

The material parameters in our simulations are those of the explosive PBX9502, but without

the slow reaction component it usually exhibits. And this is probably why we’re not able to

reproduce diameter effect data exactly.

The rod radius in our simulations is between 5 and 10 mm, and the plate thickness is between

5 and 12 mm. We show the simulation results in Fig. 1.