maintenance scheduling tool in the oil & gas industry

Maintenance Scheduling Tool in the Oil & Gas Industry

2016 Anylogic Conference

Jonatan Casiet; J.Pablo Rodriguez Varela;Patricio Pipp; Marina Pérez Gaido

Continente Siete (C7 S.A.) is an algorithm workshop,where mathematical models are constantly being developed to address complex business problems.

YPF S.A. is the largest oil & gas company in Argentina and the third in South America.

Simcastia is C7’s Simulation and Optimization division.

Who we are

Business processes analyzed:

• Wells and facilities operation

• Wells and facilities maintenance

• Well services with pulling rig

Value stream mapping

methodology:

Identification and

prioritization of issues for each

process:

Ishikawa diagram for

identification of root causes:

Brainstorming for improvement

opportunities:

Clustering:

Mess mapping; inter

relationships between clusters of improvement

opportunities

Critical businessProcess selection

Processmapping

Identification ofImprovement opportunities

Root causeanalysis

Clustering ofImprovementopportunities

Prioritization ofImprovementopportunities

Once upon a time…

• “Rincón de los Sauces” is anoil field located in Neuquén,Argentina.

• Aprox. 700 wells (amongwater injection and oil wells)

• About 100 working crews.• More than 100 weekly

maintenance orders.

Pilot location

Proposed solution

Flexibility

Is needed to develop a fully

customized tool

Optimization

Is core to the solution in order to drive objective

efficiency

Multi-paradigmSimulation

Allows the tool to be better accepted

Eye-catchingInterfaces

Help the tool to be better accepted

Cost = Resources Utilization + Wells′ Prod Loss + Dist covered

𝑓𝑓𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 × 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑓𝑓𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑑𝑑𝑑𝑑𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑑𝑑 × 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝐶𝐶𝐶𝐶𝑀𝑀𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐻𝐻𝑀𝑀𝐶𝐶𝐶𝐶𝑀𝑀+ 𝑓𝑓𝑛𝑛𝑒𝑒𝑑𝑑𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑑𝑑 × 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝐶𝐶𝐶𝐶𝑀𝑀𝐶𝐶𝐶𝐶 𝐻𝐻𝑀𝑀𝐶𝐶𝐶𝐶𝑀𝑀

𝑓𝑓𝑛𝑛𝑛𝑛𝑛𝑛𝑑𝑑 × (𝑈𝑈𝑀𝑀𝑈𝑈𝐶𝐶𝐶𝐶𝑀𝑀𝑀𝑀𝐶𝐶𝑈𝑈 +𝑃𝑃𝐶𝐶𝑀𝑀𝑀𝑀𝐶𝐶𝐶𝐶𝑃𝑃𝑃𝑃𝐶𝐶𝑈𝑈) 𝐼𝐼𝑀𝑀𝑀𝑀𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑈𝑈𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀

Cost function

Skills: type of tasks for each resource.

Parameters: Working hours, maximum extra hours.Resources

Georeferenced map: distances in between each point.

Status: conditions in the point to be maintained.Positions

Preventive & predictive: they can be planned.

Corrective: after fail detection.Tasks

Agent-based focus

Resources

Position

Tasks1. Preparation

(30’)

2. Functional Verification

(60’)

3. Maintenance (90’)

4. Sign off (30’)

Lunch

Lunch

Lunch

Work order complexity

Position

Tasks1. Preparation

(30’)

2. Functional Verification

(60’)

3. Maintenance (90’)

4. Sign off (30’)

Work order complexity

NEW AGENT REQUIRED: WORK ORDER

Scheduling Methodology

PriorityLocation

Aging

Work Order Characteristics

SequenceSimultaneityDuration

Task Characteristics

1. Order priority: lists the work order according to characteristics.2. Greedy: Analyzes possible day/time for each operation.3. Final iteration per Work Order: Adjusts tasks to minimize work order duration.

Optimization Process

1.2.3.

KRON Screens

Scheduler integration

Other Data Sources

automated input feed

automated output feed

InternalDatabase

OptimizationEngine

Cool UserInterface

MultipleUsers

local export

KRON Results

0

200

400

600

800

1000

1200

1400

1600

sep-14 oct-14 nov-14 dic-14 ene-15 feb-15 mar-15 abr-15 may-15 jun-15 jul-15 ago-15 sep-15 oct-15 nov-15 dic-15 ene-16 feb-16 mar-16 abr-16

+11%After pilot program

was initiated.

PilotProgram

Work orderexecution

Work order execution increased 11%

7377

8591 93 95 97 97 98 98 98 98 98 98

0

20

40

60

80

100

120

янв.15 фев.15 мар.15 апр.15 май.15 июн.15 июл.15 авг.15 сен.15 окт.15 ноя.15 дек.15 янв.16 фев.16 мар.16 апр.16

+25%

Upstream YPF Avg 84%

In just six months…

With just one scheduler…

With just one tool

Preventivemaintenance

Improved 25%Pilot

Program

Preventive maintenance improvement

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1

Prom2014

ene-15 feb-15 mar-15 abr-15 may-15 jun-15 jul-15 ago-15 sep-15 oct-15 nov-15 dic-15 ene-16 feb-16 mar-16 abr-16

- 56%

CorrectiveMaintenance

backlog

PilotProgram

-56%Reduction from avg

2014

Backlog reduction

0,0

5,0

10,0

15,0

20,0

25,0

30,0

35,0

40,0

Avg2013

Avg2014

abr-15 may-15 jun-15 jul-15 ago-15 sep-15 oct-15 nov-15 dic-15 ene-16 feb-16 mar-16 abr-16

- 50%

-50%Reduction from avg

2013 and 2014

Oil productionLosses due tomechanical

failures

PilotProgram

Downtime due to mechanical failures was improved

73; 80 98; 97

0

20

40

60

80

100

120

40 50 60 70 80 90 100 110

% P

reve

ntiv

e m

aint

enan

ce e

nfor

ceab

le b

y la

w

% Total preventive maintenance

AssetBPP

AssetAPP

Rincón de los sauces asset before and after the pilot program

18 MMUSD per yearAt Rincón de los Sauces Asset

NPV@12%: 234 MMUSDWith this project implemented in all the YPF assets

Economical impact

Final words

Thanks for your time!

Suggestions? Questions?

Jonatan Casiet; J.Pablo Rodriguez VarelaPatricio Pipp; Marina Pérez Gaido

Lessons learned

• Start with a pen and paper

• Do not animate just because you can

• Validation at each turn

• Optimal is great, but better is good enough