An­drea Perez-Sobers

Se­nior Re­porter

an­drea.perez-sobers@guardian.co.tt

Ar­ti­fi­cial In­tel­li­gence (AI) and dig­i­tal tech­nol­o­gy can sig­nif­i­cant­ly en­hance op­er­a­tions in ma­ture oil and gas basins, where re­sources are of­ten hard­er to find due to de­ple­tion and in­creased op­er­a­tional and in­fra­struc­tur­al chal­lenges.

That’s from Stan­ley Whar­ton, CEO of Cen­tre for En­er­gy Re­sources and Dig­i­tal­i­sa­tion Tech­nolo­gies (CERDiT) and a part-time lec­tur­er in prospect eval­u­a­tion and pe­tro­le­um eco­nom­ics at the UWI, St. Au­gus­tine.

Whar­ton said ma­ture basins im­plic­it­ly sug­gest that vast amounts of ex­plo­ration da­ta and a vast knowl­edge base ex­ist, in­clud­ing lega­cy da­ta from prospect­ing and op­er­a­tional ac­tiv­i­ties over many years. This can be a boon for the in­dus­try in the ap­pli­ca­tion of AI and dig­i­tal tech­nol­o­gy.

AI and dig­i­tal tech­nol­o­gy, he said, can help in pre­dic­tive an­a­lyt­ics and the ap­pli­ca­tion of Ma­chine Learn­ing al­go­rithms to lever­age an­a­lyt­ics for de­ci­sion-mak­ing, ef­fi­cien­cy, and pro­duc­tiv­i­ty.

“Each facet of a ma­ture basin’s ac­tiv­i­ty - ex­plo­ration, de­vel­op­ment, and pro­duc­tion - can ben­e­fit from op­ti­mis­ing process­es for faster analy­sis and de­ci­sion mak­ing and for en­hanc­ing prof­itabil­i­ty. In ex­plo­ration, da­ta an­a­lyt­ics can help in iden­ti­fy­ing new plays through the au­toma­tion of sub­sur­face an­a­lyt­i­cal process­es, man­ag­ing sub­sur­face risks, drilling ef­fi­cien­cy, com­ple­tion de­sign, and re­source man­age­ment. How­ev­er, the ex­is­tence of vast amounts of da­ta is part of the boon,” Whar­ton ex­plained.

He out­lined that da­ta must be ac­ces­si­ble and must be rig­or­ous­ly cleaned be­fore con­sid­er­ing the ben­e­fits of AI and dig­i­tal tech­nol­o­gy. Whar­ton added that con­fi­den­tial­i­ty claus­es with­in com­pa­ny and reg­u­la­to­ry in­sti­tu­tions may be coun­ter­pro­duc­tive to da­ta ac­cess and quick wins in im­ple­ment­ing AI and dig­i­tal tech­nol­o­gy.

Last week, at the T&T En­er­gy Con­fer­ence held at the Hy­att Re­gency, many CEOs of en­er­gy com­pa­nies spoke about us­ing dig­i­tal tech­nolo­gies to im­prove their work time.

Asked if AI would be fea­si­ble to use dur­ing en­er­gy com­pa­nies’ drilling ex­er­cis­es, Whar­ton high­light­ed that rig time is ex­pen­sive and AI can be used to re­duce costs through drilling ef­fi­cien­cies as it would be fea­si­ble to use dur­ing en­er­gy com­pa­nies’ drilling ex­er­cis­es.

“Tra­di­tion­al­ly, drilling re­lies heav­i­ly on hu­man ex­per­tise and ex­pe­ri­ence, which can be sub­jec­tive and lim­it­ed by the driller’s knowl­edge and re­ac­tion time. AI-pow­ered sys­tems can con­tin­u­ous­ly an­a­lyze da­ta streams dur­ing drilling from var­i­ous sen­sors (down­hole, sur­face, ge­o­log­i­cal) in re­al time for pru­dent de­ci­sion mak­ing,” he said.

Pre­dic­tive main­te­nance

Un­planned equip­ment fail­ures dur­ing drilling op­er­a­tions, Whar­ton ref­er­enced, can lead to sig­nif­i­cant down­time, cost­ly re­pairs, and even safe­ty haz­ards. AI can an­a­lyze sen­sor da­ta from drilling rigs, pumps, en­gines, and oth­er equip­ment to pre­dict po­ten­tial fail­ures be­fore they oc­cur.

“Spe­cif­ic AI ap­pli­ca­tions in­clude vi­bra­tion analy­sis, tem­per­a­ture mon­i­tor­ing and flu­id analy­sis. AI sys­tems can alert main­te­nance per­son­nel to po­ten­tial equip­ment fail­ures weeks or even months in ad­vance, al­low­ing them to sched­ule main­te­nance proac­tive­ly and avoid cost­ly down­time.”

Ques­tioned on what the down­side of AI is, the CEO said there are sev­er­al and some are as­so­ci­at­ed with strat­e­gy and plan­ning.

He not­ed that one of the main down­sides is the de­ter­mi­na­tion of the scale of im­ple­men­ta­tion, iden­ti­fy­ing the pur­pose, and defin­ing when to stop test­ing and to roll out the ap­pli­ca­tion. Bring­ing the work­force on board is key.

Al­so, he said there is a lack of trans­paren­cy and ex­plain­abil­i­ty (the ‘black box’ prob­lem); hence, a prop­er de­scrip­tion of AI and what any par­tic­u­lar ap­pli­ca­tion does must be ar­tic­u­lat­ed to work­ers and stu­dents alike.

While many ad­vanced AI al­go­rithms, par­tic­u­lar­ly deep learn­ing mod­els, op­er­ate as “black box­es” and there is a lack of clar­i­ty in un­der­stand­ing spe­cif­ic steps and rea­son­ing be­hind their de­ci­sions and to mit­i­gate this lack of un­der­stand­ing, pri­or­i­ty should be giv­en to the de­vel­op­ment and use of ex­plain­able AI tech­niques, Whar­ton said.

There are al­so the risks of cy­ber­se­cu­ri­ty and Whar­ton men­tioned AI sys­tems are vul­ner­a­ble to cy­ber­at­tacks that can com­pro­mise in­tegri­ty, avail­abil­i­ty, and con­fi­den­tial­i­ty.

Ad­di­tion­al­ly, he said the en­er­gy sec­tor is a par­tic­u­lar­ly at­trac­tive tar­get due to its crit­i­cal in­fra­struc­ture, and es­sen­tial da­ta is al­so stored in da­ta cen­tres which can be vul­ner­a­ble.

As to how many lo­cal en­er­gy com­pa­nies are mak­ing use of dig­i­tal tech­nolo­gies, he said De­N­O­VO En­er­gy and Pro­man, along with in­ter­na­tion­al ma­jors like Shell, bpTT, Wood­side, EOG, and oth­ers, may al­so have AI em­bed­ded in op­er­a­tions.

He added dig­i­tal tech­nolo­gies of­fer pow­er­ful tools for both de­tect­ing oil leaks (and gas leaks) and mon­i­tor­ing/re­duc­ing green­house gas emis­sions, con­tribut­ing to a safer, more ef­fi­cient and sus­tain­able en­er­gy in­dus­try.