New US Army Re­search Lab model knocks three cen­turies of anal­y­sis down to two weeks


Choos­ing and build­ing fu­ture weapon sys­tems just got eas­ier and faster for mil­i­tary plan­ners, thanks to a con­ver­gence of legacy and new soft­ware and the might of su­per­com­put­ers at the US Army Re­search Lab­o­ra­tory (ARL). With the smart weapon end-to-end per­for­mance model (SWEEPM), de­vel­oped by the lab, Army lead­ers can more quickly de­cide a lot quicker to mod­ify or re­place ex­ist­ing weapon sys­tems. SWEEPM can model the over­all ef­fec­tive­ness of all types of mu­ni­tions through­out the en­tire tar­get en­gage­ment, from tar­get de­tec­tion through dam­age es­ti­ma­tion with a mod­u­lar Monte Carlo sim­u­la­tion.

Right now, the tool is be­ing con­sid­ered for adop­tion by the US Army Ar­ma­ment Re­search, De­vel­op­ment and En­gi­neer­ing Cen­ter in (ARDEC) Pi­catinny. Engi­neers there say the SWEEPM model has the po­ten­tial to sim­plify per­for­mance eval­u­a­tions of both guided and un­guided mu­ni­tions through the use of mod­u­lar pro­gramme de­sign and high per­for­mance com­put­ing ca­pa­bil­i­ties.

“Unique to SWEEPM is the adap­tive na­ture of the pro­gramme which al­lows an­a­lysts at ARDEC to im­ple­ment var­i­ous guid­ance, nav­i­ga­tion and con­trol (GNC) schemes in SWEEPM via mod­ules spe­cific to that part of the mu­ni­tion,” said In­grid M. Dom­broski, Com­pe­tency Man­ager in ARDEC’s Sys­tem Anal­y­sis Di­vi­sion.

“Pre­vi­ously, the process for eval­u­at­ing new GNC schemes re­quired ei­ther writ­ing new code from scratch or mod­i­fy­ing ex­ist­ing legacy codes to fit a par­tic­u­lar need, both of which are te­dious and labour-in­ten­sive en­deav­ours”, she said.

SWEEPM is also be­ing used in a study re­quested by the Ma­neu­ver Cen­ter of Ex­cel­lence at Fort Ben­ning, Ge­or­gia, to look at per­for­mance vari­ables for the 40mm grenade.

With SWEEPM, ARL ex­perts can pass along per­for­mance in­di­ca­tors about the 40mm grenade in about two weeks, but with­out the model and the use of su­per­com­put­ers, it would take re­search an­a­lysts 316 years to per­form such anal­y­sis.

“The model is a set of files and soft­ware that can be eas­ily pack­aged and shared across a file share net­work or burned to a disk”, ex­plained Mary K. Arthur, the prin­ci­ple in­ves­ti­ga­tor who is cred­ited with de­vel­op­ing SWEEPM af­ter recog­nis­ing a void in mod­el­ling and analysing smart weapon sys­tems from tar­get ac­qui­si­tion through dam­age es­ti­ma­tion. She is a math­e­ma­ti­cian in the Lethal­ity Di­vi­sion of ARL’s Weapons and Ma­te­ri­als Re­search Direc­torate. SWEEPM was con­ceived in 2008 and com­pleted in April 2013. “Users have a cou­ple of op­tions as to what they would get. First, they can choose be­tween the Linux and the Win­dows ver­sions, which only dif­fer in that the Linux ver­sion sup­ports par­al­lel pro­cess­ing. They also have the op­tion of re­ceiv­ing just the ex­e­cutable, or re­ceiv­ing the en­tire source code and nec­es­sary files and di­rec­tions for com­pil­ing on their own ma­chines.”

As a plat­form in­de­pen­dent code, it can be run as a sin­gle Monte Carlo sim­u­la­tion or spawned off as par­al­lel pro­cesses al­low­ing for more com­plex physics-based fea­tures to be em­ployed through­out the en­gage­ment sce­nario.

Em­bed­ded sim­u­la­tions in­clude mul­ti­ple fir­ing plat­forms, scouts, mov­ing tar­gets and col­lab­o­ra­tive smart pro­jec­tiles to model both be­yond line of sight and non-line of slight smart weapon sys­tems, where tar­get­ing data is passed from the scout ve­hi­cle to re­mote fir­ing plat­forms.

The Op­er­a­tional Re­quire­ment-based Ca­su­alty As­sess­ment (ORCA) mod­el­ling sys­tem, cre­ated by ARL’s Sur­viv­abil­ity/Lethal­ity Anal­y­sis Direc­torate in 1996, is also em­bed­ded in SWEEPM. ORCA was de­vel­oped for tri-ser­vice to al­low as­sess­ments of Sol­dier per­for­mance fol­low­ing weapon-in­duced in­jury.

It per­mits ca­su­alty as­sess­ments to be per­formed in a con­sis­tent man­ner across vir­tu­ally all types of mil­i­tary plat­forms, jobs and weapon-in­duced threats. The ORCA mod­el­ling sys­tem in­cor­po­rates pre­vi­ously de­vel­oped as well as newly de­vel­oped in­jury cri­te­ria mod­els, al­go­rithms and scor­ing sys­tems to char­ac­terise hu­man bio-re­sponse to trauma from var­i­ous types of bat­tle­field in­sults and de­rives es­ti­mates of Sol­dier per­for­mance degra­da­tion.

Re­searchers rely on high per­for­mance com­put­ers (HPC), housed within the ARL Su­per­com­put­ing Re­search Cen­ter man­aged by ARL’S Com­pu­ta­tional and In­for­ma­tion Sciences Direc­torate at Aberdeen Prov­ing Ground. SWEEPM’s high-fidelity physics sub­mod­ules can only be run on th­ese su­per­com­put­ers.

Put in per­spec­tive, a sin­gle it­er­a­tion in a Monte Carlo sim­u­la­tion takes one sec­ond and, “Let’s say that you want to per­form a 10,000 it­er­a­tion sim­u­la­tion, then that sim­u­la­tion would take roughly three hours,” Arthur ex­plained. “Not too bad. How­ever, say that a sin­gle it­er­a­tion takes 20 sec­onds. That same sim­u­la­tion now takes closer to 2.5 days to com­plete”.

“An en­gage­ment sce­nario with 10 mov­ing tar­gets and 10 guid­ing pro­jec­tiles can eas­ily take 20 sec­onds if not much more. Fi­nally, say you need 1,00,000 it­er­a­tions for good con­ver­gence of the data. The sim­u­la­tion will now take over 7.5 months to run to com­ple­tion.”

For the 40mm grenade study, ARL re­search an­a­lysts want to per­form 10,00,000 sim­u­la­tions of 10,000 it­er­a­tions each.

“I cal­cu­late that it would take 316 years to per­form this anal­y­sis in se­rial even with a de­cent pro­ces­sor. By tak­ing ad­van­tage of the HPC, I am con­fi­dent that we can cut the run­time down to un­der two weeks as a con­ser­va­tive es­ti­mate,” said Arthur.

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.