New Technology for Dramatically Improving Reciprocating Compressor Performance

New Technology for Dramatically Improving Reciprocating Compressor Performance Performance Augmentation Network Technology (PAN)

Brief Technical Background 2014: Sta. 85 field installation & testing. GMRC Paper on test results. 2013: Commercial release of Virtual Pumping Station & Design Software; Project opportunity at Williams sta. 85. 2012: Kinder Morgan acquired El Paso & withdrew Batesville. 2011: Research to increase compressor efficiency via tuning; GMC Paper on Performance Augmentation Networks; El Paso Batesville Sta. design & GMRC project. initiated. 2010: Designed PAN for El Paso Sta. 96; Project cancelled due changing pipeline flows. Continued research on PAN tuned manifolds. 2009: 2 throw proof of concept test on Superior MH64 at TGT Ellisburg station; GMC Paper comparing PAN simulation & field test results. 2008: Lab testing of PAN for 2 DA cyl. air compressor; GMC Paper on Air Compressor Test Rig 2007: OPT & ACI cooperate to model a cylinder compressor; GMC Paper on Tuned Loops and Tuning Section Transitions 2006: OPT & El Paso cooperate to model Ariel compressor cylinder.

The History of Unsteady Gas Dynamics in Pipes Initial development in Europe in the mid 20 th Century. Queens University of Belfast took the academic research lead. Professor Gordon P. Blair, CBE became its leading spokesperson publishing over 100 SAE papers and two reference text books. High performance racing engine simulation and design was his passion.

The Development of Virtual Engines and Virtual Pumping Stations 1993 OPTIMUM Power Technology entered into a development partnership with QUB and developed its Virtual Engines design software based on this physics. VE was licensed to Honda, Toyota Racing Development (TRD), Ilmor, Hendrick Motorsports and others who used it very successfully for racing engine design. VE was licensed to Cameron for natural gas engine development. Because engines breathe air at 1 atmosphere, the thermodynamics of VE is based on ideal gas laws. OPTIMUM Pumping Technology enhanced VE to accurately simulate natural gas reciprocating compressors Virtual Pumping Station. Virtual Pumping Station was enhanced to use real gas physics based on NIST data. Dynamic (mass spring) check valve support was added. Infinite step unloader support. Loop PAN pulsation comb filter technology was developed. Manifold PAN technology was developed. Cameron became the 1 st VPS customer.

Principles of Unsteady Gas Dynamics in Pipes Finite amplitude unsteady gas dynamics is fundamentally different from and far more complex than steady state gas flow theory. It is as different as electrical AC circuit theory is from DC circuit theory. Finite amplitude unsteady gas dynamics is fundamentally different from and far more complex than acoustic theory. Finite amplitude pressure waves change the media as they propagate. They transmit and reflect at pipe area changes and everywhere the media in the pipe changes its properties. In any pipe that is experiencing unsteady-pulsing flow, there are 2 waves propagating in opposite directions. The convention is to call them the Right wave and the Left wave. The sum of these 2 waves is the superposition pressure that we measure with pressure transducers.

PAN Virtual Pumping Station Animation of Cylinder 1 Head End, Crank End, Check Valves and Header Pipe

PAN Virtual Pumping Station Flow Diagram

Bottles and PANs are FUNDAMENTALLY Different Bottle systems Reduce pulsation to acceptable levels DISSIPATE 95% of pulsation energy Process each throw INDIVIDUALLY Components Primary and secondary bottle volumes Choke tubes Orifice plates Problems: Pressure loss Poorly phased cyclical flange pressure that reduces adiabatic efficiency of the compressor Bottle vibration problems Bottles CANNOT be modified to perform like a PAN PANs Reduce pulsation to acceptable levels RECOVER 95% of pulsation energy Process ALL throws as a SYSTEM Components Primary TST W or Y junction recovers pulsation energy Secondary Y TST junction to join flow from both sides of compressor to reduce pulsation Standard pipe OPTIMIZED TSTs, pipe diameters and lengths Benefits: NO pressure loss Greatly improved cyclical flange pressure significantly increases compressor adiabatic efficiency Reduced operation cost Reduced power/engine emissions Increased flow Reduced vibration caused failures

2008: Laboratory (2) DA Cylinder Air Test 2 x Quincy single stage 500 1000 rpm 2 DA cylinder discharge PAN Pressure (PSI) 26.5 1000 RPM, 21 PSIA, Y with 15' & 10' loops 26 Full load 25.5 25 24.5 24 23.5 23 22.5 22 21.5 21 0.7% P/P discharge pulsation 20.5 20 Head 1 Head 2 After Y After 1st loop After 2nd loop 19.5 19 18.5 18 17.5-180 -135-90 RECIPROCATING -45 0 COMPRESSOR 45 90 PULSATIONS 135 180 Crank angle (deg.)

2009: El Paso Ellisburg Field Test 2 nd Generation Superior MH64 single stage 750 1000 rpm 2 DA cylinder discharge PAN 11 10 9 PAN Side vs BOTTLE Side Pressure Drop Comparison Pressure drop (PR = 1.1) 896 psig Avg. Discharge Pressure 940.0 930.0 1000 RPM Fully Loaded Pressure drop (PSI) 8 7 6 5 4 Cyl. 3 Cyl. 1 Traditional Bottle System - Cyl. 1 & 3 920.0 910.0 5.5 psig less discharge pressure drop 0.4% P/P discharge pulsation Pressure (PSI) 900.0 3 2 1 Cyl. 4 Cyl. 2 Bottle-less Pan System - Cyl. 2 & 4 890.0 880.0 Nozzle Start of 1st loop After PAN 0 860 880 900 920 940 960 980 1000 Speed (RPM) BOT Cyl 1 to LINE PAN Cyl 2 to LINE BOT Cyl 3 to LINE PAN Cyl 4 to LINE 870.0 Linear (PAN Cyl 2 to LINE) Linear (PAN Cyl 4 to LINE) Linear (BOT Cyl 1 to LINE) Linear (BOT Cyl 3 to LINE) 0 45 90 135 180 225 270 315 360 405 450 495 540 585 630 675 720 Angle

2010: El Paso Sta. 96 Conversion Design 3 rd Generation Ariel KVB/6 single stage 720 rpm constant speed 6 DA cylinder PAN manifold Infinite step unloading Cancelled due to PL flow demand change. Flange > W TST > Side Pipe > Y TST > Pipeline

2011: El Paso Batesville Conversion Design 4 th Generation Ariel KVB/6 single stage 500 750 rpm 6 DA cylinder PAN manifold 8 symmetric unloading steps Cancelled due to KM acquisition. GMRC & El Paso Project Objectives No Compressor Modifications Variable Speed 8 Load Steps with ACI Pockets CONTINUOUS Unloading Less than 1% P/P pulsation under all operating conditions 90% Reduction in system pressure Loss Acceptable vibration Demonstrate that PANs can significantly improve the efficiency far beyond the benefits of reduced pressure loss

4 th Generation PAN Improved Adiabatic Efficiency

2013 Progress & Status GMRC Royalty Agreements have been negotiated with OPTIMUM Pumping A second Virtual Pumping Station 6 throw baseline bottle compressor model has been 90% validated New test site at Williams Sta. 85 identified May 2013 2 identical units to be re cylindered: one with PAN; one with new bottles. Project schedule is fast track and has to move forward right away to meet the station s contractual needs. Project has been planned to accomplish GMRC Project s objectives, with sequence adapted from El Paso project. Modeling & simulation completed and presented herein. Mechanical design nearly completed. System mechanical analysis ready to start after review meeting. TST production ready to start. Williams project team $$ approval pending (now the pacing item in schedule). 2 identical Cat G3616 & Ariel JGZ/6 Packages Requiring Re cylindering 1 unit to have PAN 1 unit to have new bottles

Williams Sta. 85 New Cylinders & Unloading Cat G3616 & Ariel JGZ/6 with (6) 9.75 ZL cylinders Unloading selected based on bottle system. HE FVCPs on each cylinder. No CE FVCPs required. No SACE load steps required.

Primary Project Objectives Total system pressure drop from suction line to discharge line less than 2 psig at all operating points. Control of pulsation to less than 1.5% of line pressure level at all operating conditions. Control of mechanical vibrations and stress levels consistent with API 618 M5 requirements. A 10% reduction in compressor BHP/MMSCFD at the high flow operating condition (compared to bottle system).

Secondary Project Objectives The ability of OPTIMUM s Virtual Pumping Station software to predict the pressure drop, pulsation and BHP/MMSCFD of the re cylindered compressor with the PAN installed. The ability of the Virtual Pumping Station software to predict the pressure drop, pulsation and BHP/MMSCFD of the re cylindered compressor with the Bottles installed. The ability of OPTIMUM s design optimization software to create a PAN that simultaneously achieves all of the primary objectives of the project.

Project Team Performance Augmentation Network Technology (PAN) ACI Services Inc. Norm Shade ACI (executive project management) John Bazaar ACI (project management; mechanical design, analysis) Tyler Clark ACI (mechanical design, analysis, CAD modeling) Jessica Gilcher ACI (design, CAD modeling) Joe Mosely ACI (CAD modeling) Nak Nortey ACI (FEA) Beta Machinery Analysis (system mechanical analysis) OPTIMUM Power Technology Glen Chatfield OPT(project management; simulation & PAN design) Dale Wells OPT(simulation & PAN design) Malcom Ashe OPT(valve modeling/simulation) Williams Patrick Jacobs Scott Schubring

5/12* Schedule Project Schedule with Milestones Cylinder selection (Williams) 5/15/13 Done Final cost estimate & schedule to Williams 5/31/13 Done Draft contract to GMRC 6/7/13 8/31/13 est. Initial PAN model development & simulation (OPT) 6/15/13 Done GMRC contract signed 6/30/13 10/7/13 Invoice #1 to GMRC 7/23/13 10/7/13 Begin solicitation of co funding 7/1/13 9/1/13 Mechanical layout (ACI) 6/30/13 Done Final PAN model (ACI/OPT) 7/8/13 Done Present PAN Model results to Williams & PSC 7/12/13 8/13/13 Current Williams & GMRC decision to proceed 7/16/13 8/15/13 Complete TST design & FEA (ACI) 8/7/13 9/16/13 Complete detailed mechanical design (ACI) 8/15/13 8/30/13 Finalize support & pier/foundation design (ACI) 8/15/13 8/30/13 Begin TST patterns (ACI) 8/16/13 10/1/13 Complete valve analysis 9/16/13 10/15/13 Complete system mechanical analysis (ACI/contractor) 9/16/13 10/15/13 Present mechanical analysis results to Williams & PSC 9/18/13 10/17/13 Williams & GMRC decision to proceed 9/20/13 10/19/13 Invoice #2 to GMRC 9/22/13 10/19/13 * l / / Current Schedule

5/12* Schedule Project Schedule with Milestones Current Schedule Order PAN material (ACI) 9/23/13 10/21/13 Complete new #2 unit bottle design (Enerflex) 10/31/13 10/31/13 Model & simulate new bottles (OPT) 11/30/13 11/30/13 Complete test protocol & define instrumentation N/A 12/15/13 Invoice #3 to GMRC 1/7/14 1/7/14 Complete TSTs (ACI) 1/15/14 1/15/14 Complete PAN & supports fabrication (ACI/contractor) 2/28/14 2/28/14 Invoice #4 to GMRC 3/1/14 3/1/14 Remove existing bottle system & install piers (Williams) 3/31/14 3/31/14 Install new cylinders (Williams) 4/8/14 4/8/14 Install PAN on Unit #1 & bottles on Unit #2 (Williams) 4/22/14 4/22/14 Hydrostatic testing and support installation (Williams) 4/30/14 4/30/14 Onsite static mechanical test (team) 5/7/14 5/7/14 Mechanical & performance testing (ACI/OPT/contractor) 5/24/14 5/24/14 Acceptance 5/31/14 5/31/14 Invoice #5 to GMRC 6/14/14 6/14/14 Final report to Williams & GMRC 8/?/14 8/12/14 Invoice #6 to GMRC 8/31/14 8/12/14 Report results at 2014 GMC 10/?/14 10/6/14 * Preliminary estimate 5/12/13

Mechanical Design As Is Performance Augmentation Network Technology (PAN)

Mechanical Design 5 th Generation PAN 8 8 12 12 16

Mechanical Design Preliminary Supports & Platform

Mechanical Design Top View

Mechanical Design Top View

Mechanical Design Site Work

Mechanical Design Site Work

Mechanical Design Site Work

Mechanical Design Site Work

Mechanical Design Prefabrication Suct W-TST Suct Y-TST Disch W-TST Disch Y-TST

Performance Prediction Suction Pressure vs. Crank Angle @ 950 RPM Design Pt. 3: 850 psig (864.7 psia) suction / 1307 psig (1321.7 psia) discharge fully loaded @ Cyl. Flange After W-TST After Y-TST

Performance Prediction Suction Pressure Animation @950rpm Design Pt. 3: 850psig (864.7 psia) suction / 1307psig (1321.7 psia) discharge fully loaded

Performance Prediction Discharge Pressure vs. Crank Angle @ 950 RPM Design Pt. 3: 850 psig (864.7 psia) suction / 1307 psig (1321.7 psia) discharge fully loaded @ Cyl. Flange After W-TST After Y-TST

Performance Prediction Discharge Pressure Animation @950rpm Design Pt. 3: 850psig (864.7 psia) suction / 1307psig (1321.7 psia) discharge fully loaded

Performance Prediction PV Chart @950rpm Design Pt. 3: 850psig (864.7 psia) suction / 1307psig (1321.7 psia) discharge fully loaded

Performance Comparison Performance Augmentation Network Technology (PAN) STATION 85 6 VERY DIFFERENT DESIGN POINT OPERATING CONDITIONS Comparison of ercm Bottle** predictions to VPS PAN predictions PAN Averages 17% Greater Efficiency and PAN Averages 14% Greater Flow Never Exceeded Rated BHP at Full Load (i.e., No Unloading) ** Bottle system pressure drop 2% suction and 1% discharge (lower than GMRC Guideline).

Performance Prediction BHP v. RPM Design Pt. 3: 850 psig suction / 1307 psig discharge fully loaded

Performance Prediction MMSCFD v. RPM Design Pt. 3: 850 psig suction / 1307 psig discharge fully loaded

Performance Prediction BHP/MMSCFD v. RPM Design Pt. 3: 850 psig suction / 1307 psig discharge fully loaded Very good over entire speed range.

Performance Prediction Suction P/P Pulsation v. RPM Design Pt. 3: 850 psig suction / 1307 psig discharge fully loaded 0.1 to 0.5%

Performance Prediction Discharge P/P Pulsation v. RPM Design Pt. 3: 850 psig suction / 1307 psig discharge fully loaded 0.1 to 0.6%

Performance Prediction Suction Pressure Loss v. RPM Design Pt. 3: 850 psig suction / 1307 psig discharge fully loaded 0.4 to 0.2 psi pressure gain

Performance Prediction Discharge Pressure Loss v. RPM Design Pt. 3: 850 psig suction / 1307 psig discharge fully loaded 0.0 to 0.5 psi pressure loss [Total avg. Suct + Disch Pressure Loss = 0.06 psi]

Performance Prediction BHP v. RPM All Design Points fully loaded Performance Augmentation Network Technology (PAN)

Performance Prediction MMSCFD v. RPM All Design Points fully loaded Performance Augmentation Network Technology (PAN)

Performance Prediction BHP/MMSCFD v. RPM All Design Points fully loaded Very good over entire speed range. Performance Augmentation Network Technology (PAN)

Performance Prediction Suction P/P Pulsation v. RPM All Design Points fully loaded 0.2 to 0.7% Performance Augmentation Network Technology (PAN)

Performance Prediction Discharge P/P Pulsation v. RPM All Design Points fully loaded 0.1 to 0.6% Performance Augmentation Network Technology (PAN)

Performance Prediction Suction Pressure Loss v. RPM All Design Points fully loaded 0.4 to 0.1 psi pressure gain Performance Augmentation Network Technology (PAN)

Performance Prediction Discharge Pressure Loss v. RPM All Design Points fully loaded 0.6 to +0.6 psi pressure loss Performance Augmentation Network Technology (PAN)

Performance Prediction BHP v. RPM All Design Points fully unloaded (6 HE pockets open)

Performance Prediction MMSCFD v. RPM All Design Points fully unloaded (6 HE pockets open)

Performance Prediction BHP/MMSCFD v. RPM All Design Points fully unloaded (6 HE pockets open) Very good over entire speed range.

Performance Prediction Suction P/P Pulsation v. RPM All Design Points fully unloaded (6 HE pockets open) 0.6 to 1.3%

Performance Prediction Discharge P/P Pulsation v. RPM All Design Points fully unloaded (6 HE pockets open) 0.3 to 1.2%

Performance Prediction Suction Pressure Loss v. RPM All Design Points fully unloaded (6 HE pockets open) 0.3 to 0.0 psi pressure gain

Performance Prediction Discharge Pressure Loss v. RPM All Design Points fully unloaded (6 HE pockets open) 0.0 to 0.5 psi pressure loss

Primary Project Objectives Comparison Total system pressure drop from suction line to discharge line less than 2 psi at all operating points. Simulation Results: All points less than 0.3 psi = Objective met. Control of pulsation to less than 1.5% of line pressure level at all operating conditions. Simulation Results: Worst case 0.6% fully loaded / 1.3% with maximum unloading = Objective met. Control of mechanical vibrations and stress levels consistent with API 618 M5 requirements. Next investigative phase (to be reported in October) A 10% reduction in compressor BHP/MMSCFD at the high flow operating condition (compared to bottle system). Simulation Results: average 17% reduction at 6 design points = Objective met.

Project Cost Estimate Cost Performance Augmentation Network Technology (PAN) Estimated Cost ACI/OPT Contribut n Williams Cost* GMRC Cost System Modeling, Optimization & Eng. (OPT) $150,000 $0 $0 $150,000 Previous uncompensated Batesville Sta. (OPT) $95,000 $95,000 $0 $0 Previous uncompensated Batesville Sta. (ACI) $67,000 $67,000 $0 $0 System Design and Drwg Development (ACI) $72,000 $72,000 $0 $0 System Mechanical Analysis (contractor) $27,500 $0 $0 $27,500 Project Management Services (ACI ) $32,000 $32,000 $0 $0 TSTs including P&T for new designs (ACI) 229,200 $71,500 $0 $157,700 Remove Existing System (Williams) $28,300 $0 $28,300 $0 Procure/Fabricate PAN & supports (ACI) 283,250 $0 $0 $283,250 Install New PAN System (Williams) 298,700 $0 $298,700 $0 Site Engineering Support (ACI & OPT) $96,000 $96,000 $0 $0 Field Mech & Perf Testing/Analys (contractor) $55,000 $0 TBD $55,000 Travel Expenses for site support (ACI and OPT) $66,550 $0 $0 $66,550 Contingency Modifications $100,000 $0 $0 $100,000 PROJECT TOTAL $1,650,500 NEW TECHNOLOGY $433,500 $327,000 FOR CONTROLLING $840,000 *Approval Pending

Project Cost Estimate Performance Augmentation Network Technology (PAN) Invoice Schedule to GMRC Amount Date GMRC Invoice 1 15% $126,000 10/7/13 GMRC Invoice 2 10% $84,000 10/19/13 GMRC Invoice 3 25% $210,000 1/7/14 GMRC Invoice 4 20% $168,000 3/1/14 GMRC Invoice 5 25% $210,000 6/1/14 GMRC Invoice 6 5% $42,000 8/31/14 GMRC TOTAL $840,000 GMRC Funding Sources & Schedule GMRC approved maximum 2012 $250,000 Available Ariel Co funding $100,000 Available Caterpillar Co funding $100,000 Available GE Water & Power Co funding $50,000 Available Dresser Rand Co funding $20,000 Available Additional GMRC request 2013 $150,000 10/11/13 Additional Co funding TBD $170,000 3/31/13 GMRC TOTAL $840,000 Contingency Co funding TBD $100,000