Enhanced Procurement and Production Strategies ... - ACS Publications

Subscriber access provided by WEBSTER UNIV

Process Systems Engineering

Enhanced Procurement & Production Strategies for Chemical Plants: Utilising Real-Time Financial Data and Advanced Algorithms Janusz Jerzy Sikorski, Oliver R Inderwildi, Mei Qi Lim, Sushant Suhas Garud, Johannes Neukäufer, and Markus Kraft Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.8b02925 • Publication Date (Web): 05 Feb 2019 Downloaded from http://pubs.acs.org on February 6, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Enhanced Procurement & Production Strategies for Chemical Plants: Utilising Real-Time Financial Data and Advanced Algorithms

Janusz J. Sikorski,

†,‡

‡

Oliver Inderwildi,

‡

Johannes Neukäufer,

Mei Qi Lim,

¶,‡

§,‡

Sushant S. Garud,

∗,†,¶,‡

and Markus Kraft

†Department of Chemical Engineering and Biotechnology, University of Cambridge, United

Kingdom ‡CARES, CREATE Tower, 1 Create Way 138602, Singapore ¶School of Chemical and Biomedical Engineering, Nanyang Technological University,

Singapore §Department of Chemical and Biomolecular Engineering, National University of Singapore,

Singapore E-mail: [email protected]

Phone: +44 (0)1223 762784

Abstract This paper presents an implementation of an automated algorithm powered by market and physical data to improve procurement & production of a chemical plant with the goal of improving the overall economics on the entity. Herein, the algorithm is applied to two scenarios which serve as case studies: conversion of natural gas to methanol and crude palm oil to biodiesel. The programme anticipates opportunities to increase prot or avoid loss by analysing the futures market prices for both reagents and the products 1

ACS Paragon Plus Environment

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

while considering cost of storage and conversion derived from physical simulations of the chemical process. Analysis conducted on 11.06.2018 in the biodiesel scenario shows that up to 219.28 USD per tonne of biodiesel can be earned by buying contracts for delivery of crude palm oil in July 2018 and selling contracts for delivery of biodiesel in August 2018 which equates to a margin 11.6% higher than in case of the direct trade. Moreover, it is shown that losses of up to 11.3% can be avoided and therefore, it is shown that there is realistic scope for increasing the protability of a chemical plant by exploiting the opportunities across dierent commodity markets in an automated manner. Consequently, such a cyber system can be used to assist eco-industrial parks with supply-chain management, production planning as well as nancial risk governance and, in the end, help to establish a long-term strategy. This study is part of a holistic endeavour that applies cyber-physical systems to optimise eco-industrial parks so that energy use and emissions are minimised while economic output is maximised.

Keywords: big data, machine intelligence, commodities, sustainability, cyber-physical system, futures market, articial intelligence, sustainability, optimisation

Introduction Most organisations participating in industrial parks are prot-seeking companies. Sometimes investment and energy-saving opportunities present themselves, but cannot be exploited due to signicant risks and absence of relevant data or a physical model with suciently strong predictive capabilities. For example, a chemical plant, with spare production capacity, could make additional earnings by exploiting the price dierences between its feedstock and product on the commodity markets. However, straying from the established throughput without accurate simulations based on enhanced algorithms and articial intelligence may lead to decreased product quality or even damaged equipment. Industry 4.0 introduces many concepts relevant to taking advantage of the aforemen2


Page 2 of 28

Page 3 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


tioned opportunities, including machine-to-machine (M2M) communication, cyber-physical systems (CPSs) and the Internet of Things (IoT) 1,2 . M2M communication refers to the ability of industrial components to communicate with each other. CPSs can monitor physical processes, create virtual copies of the physical world and make decentralised decisions thereby acting as AI supporting human decision making. IoT is a dynamic network where physical and virtual entities have identities and attributes and use intelligent interfaces. A system could be established that gathers and analyses market (local and global) and physical (from sensors and simulations) data about the relevant industrial processes and advises on potential investments and energy savings. It is conceivable that in the future an entire production plant or even an entire industrial park would autonomously seek and fulll such opportunities. Furthermore, exploring and exploiting such possibilities within an industrial park may encourage closer cooperation of participants' plants leading to a transformation into an ecoindustrial park (EIP). An EIP is an industrial park where businesses cooperate with each other and, at times, with the local community to reduce waste and pollution, eciently share resources (such as information, materials, water, energy, infrastructure, and natural resources), and minimize environmental impact while simultaneously increasing business success 35 . Numerous studies have been written on trading commodities and their futures on an exchange, for example, Garcia and Leuthold 6 , Szakmary et al. 7 , Fung and Hsieh 8 , Campa 9 . A number of publication considers the interactions between the production and the commodity markets. Smith and Stulz 10 examine the reasons why rms hedge and what risks do they choose to hedge and develop a theory of the hedging behavior of value-maximizing corporations. Bjorgan et al. 11 study the issues of nancial risk management in the energy sector and explores impact of nancial contracts (such as futures) for scheduling policies of the companies in the industry. The paper by Tanlapco et al. 12 examines risk-minimizing hedging strategies using futures contracts in an electricity market and nds that the use of 3



electricity futures contracts is superior to using other related futures contracts such as crude oil. Spinler et al. 13 present a theoretical analysis of options contracts for physical delivery which shows how spot market price risk, demand and cost risk can be shared between buyer and seller. They conclude that such a contingency may complement nancial risk management instruments capital-intensive rms such as those in the chemical industry. Ding et al. 14 explore integrated operational and nancial hedging decisions faced by a global rm. The study found that nancial hedging strategy ties closely to, and can have both quantitative and qualitative impact on, the rm's operational strategy. The publication by Kannegiesser et al. 15 presents a planning model for coordinating sales and supply decisions for commodities in a chemical industry. The impact of elasticities, variable raw material consumption rates and price uncertainties on planned prot and volumes are demonstrated. A number of studies develop and apply modelling techniques in order to aid companies with nancial risk management. Ryu 16 discusses two multi-period planning strategies used to minimise negative impact of varying conditions on a company's protability. One is to modify the external condition, e.g. demands, and the other is to expose explicit constraints limiting capacity expansion. Park et al. 17 and Ji et al. 18 present nancial risk management methods in the petrochemical industry. The former demonstrates a two-stage stochastic programming framework for operational planning and nancial risk management of a renery. The method optimised the contract sizes (long-term, spot and futures) and the plant's operational plan. The latter presents a one-stage stochastic programming model for the integration of operational hedging and nancial hedging strategy in the crude oil procurement process subject to oil price uctuation. This approach uses Conditional Value-at-Risk as the risk measure and considers futures contracts, put options and call options during optimisation. A publication by Longinidis et al. 19 presents a supply chain network design model that yields the optimal conguration under a variety of exchange rate realizations and integrates operational hedging actions that mitigate exchange rate risk. Kwon et al. 20 describe and demonstrate a two-stage programming framework for maximizing prot of a petrochemical 4


Page 4 of 28

Page 5 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


company. The two stages are hedge trading (minimising costs of raw materials) and production planning (maximising sales prots). The procedure simultaneously optimises timing, amount, and price of raw materials and the strategies for facility operation and product sales. An example of a Korean petrochemical company is used to present that this approach can improve protability. Guo et al. 21 present two parameters for investigating market eciency and presence of arbitrage opportunities or anomaly: stochastic dominance (SD) and Omega ratio. The paper studies the relationship between the parameters and applies the developed theory to examine the relationship between property size and property investment in the Hong Kong real estate market. Ng et al. 22 propose tests for stochastic dominance constructed by translating the inference problem of stochastic dominance into parameter restrictions in quantile regressions and apply them to the NASDAQ 100 and S&P500 indices to investigate dominance relationship before and after major turning points. While many studies theoretically explore trading commodities and their futures on an exchange and develop methods for optimising this activity, none among the found literature have investigated and implemented a system gathering market and physical data in order to make recommendations about and act on opportunities for saving energy and increasing protability. The purpose of this paper is to present an automated procedure powered by market and physical data applied to two scenarios: conversion of natural gas to methanol and crude palm oil to biodiesel (fatty acid methyl ester - FAME). The algorithm uses current market data as well as futures curves for starting materials and products to anticipate the largest margins to be made under the consideration of storage costs. Figure 1 illustrates the concept of the smart algorithm - it assists the business with the decision making for procurement, production and sale: Should a product be sold or stored? Should a reagent be bought and stored or directly converted? etc. Consequently, the algorithm represents articial intelligence that aids the decision making of both procurement and production planning of a chemical plant by analysing data from 5



Figure 1: Scheme of the functioning of the smart algorithm: Futures curves for commodity markets are utilised to anticipate market movements and subsequently, cost dierences are exploited to maximise margins or indeed avoid losses nancial markets. Such algorithm can therefore not only optmise the procurement but also aid the production planning to maximise prots. This paper is structured as follows: section provides design and implementation details of the price discrepancy spotter; section describes and discusses results of the analysed case studies; section summarizes the main ndings and draws conclusions from the presented case studies .

Automated price-dierence analysis Design and implementation

This section describes the design and implementation of the automated price-discrepancy spotter for a chemical plant converting a reagent into a product. The spotter searches for opportunities to make additional prot by analysing the futures market prices for both the reagent and the product. It considers costs of storage, transport and conversion (steam, electricity and other utilities) and produces recommendations for procurement & production 6


Page 6 of 28

Page 7 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


to optimise the overall protability of the plant. It is assumed that the plant is located in Singapore and operates with a long-term production contract. The implementation involves the following steps: 1. downloading market data from an appropriate online source (the considered scenarios include the Chicago Mercantile Exchange 23 and Zhengzhou Commodity Exchange 24 ), 2. performing a feasibility analysis based on the downloaded data and physical data provided by a simulation of the chemical plant under consideration (the included scenarios employ Aspen Plus 25 and Aspen HYSYS 26 ). Both steps are executed by scripts written in Python 3.5.0 27 . Market data is downloaded on a daily basis as it is updated at the same rate. The second step determines potential prot to be made on delivering a single unit of the product. The required quantities of the reagent contracts, utilities (steam, water, heating fuel and electricity), transport to and from the plant and storage volume are calculated and costed. The utility, transport and storage prices were determined based on specications of a biodiesel plant designed by Lurgi GmbH in 2007 and the available literature. Necessary currency conversions were done with data from Oanda 28 , XE.com 29 , while Coinnews Media Group LLC 30 was used to adjust for ination. The programme solves the following equation for all contracts in the correct chronological order (i.e. delivery of the product must be scheduled at least a month after delivery of the reagent) in order to determine prot per unit of the product:

Pp − Tp − (Sp ⊕ (Sr × R)) × d − R × (Pr − Tr ) − U

(1)

where subscripts p and r refer to the product and the reagent respectively, P to price, T to transport cost, S to storage cost, R to units of the reagent per unit of the product, d to the storage duration and U to the cost of utilities per unit of the product. Note that ⊕ indicates that the programme assumes either the product or the reagent are stored for the entire time 7



between reception of latter and delivery of the former. In the considered scenarios P was taken from Chicago Mercantile Exchange 23 and Zhengzhou Commodity Exchange 24 and T ,

R, S and U were calculated using the data from Tables 2 and 5 and Aspen Plus and Aspen HYSYS plant simulations. Finally, the software will return the most protable contract and storage schedule combination.

Software Aspen Plus V8.8

Aspen Plus 25 is a process modelling and optimisation software used by the bulk, ne, specialty, and biochemical industries, as well as the polymer industry for the design, operation, and optimisation of safe, protable manufacturing facilities. Its capabilities include: optimisation of processing capacity and operating conditions, assessment of model accuracy, monitoring safety and operational issues, identifying energy savings opportunities and reduce greenhouse gas (GHG) emissions, performing economic evaluation, improving equipment design and performance.

Aspen HYSYS V8.8

Aspen HYSYS 26 is a process modeling tool used by the petrochemical industry for process simulation and process optimization in design and operations. Its uses include, but are not limited to: designing various components in a petrochemical plant, 8


Page 8 of 28

Page 9 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


detecting abnormal operating conditions, modeling steady state and dynamic processes.

Python 3.5.0

Python 27 is a high-level, interpreted programming language for general-purpose programming. It emphasizes readability and an expressive syntax. In this study it was used to download online data, perform calculations and connect to Aspen Plus and Aspen HYSYS. The last activity was performed using Microsoft Component Object Model (COM) interface which is a platform-independent, binary-interface standard enabling creation of objects and communication between them 31 . COM object (also known as COM component) is dened as a piece of compiled code that provides a service to the rest of the system.

Results and discussion Crude palm oil-to-biodiesel conversion Biodiesel plant simulation

A process producing 24.334 tonnes per hour of biodiesel was modelled in Aspen Plus V8.8 using the UNIFAC-DMD property method 32 (including steam tables for calculations involving pure water). The ow sheet model includes two reaction stages (modelled using continuously stirred tank reactors - CSTRs), a separation stage, a methanol recycle loop, a gas-fuelled steam generation section and auxiliary equipment, see Fig. 2. The nal fuel, fatty acid methyl ester, is produced via trans-esterication pathway where triglycerides react with methanol to form methyl ester and glycerol in the presence of an alkaline catalyst. The ow sheet was based on an existing plant designed by Lurgi GmbH. In the process tripalmitine oil is reacted with methanol in the CSTRs to produce glycerol and methylpalmitate (biodiesel), then passed through a ash drum and a decanter and washed with water to separate the 9



remaining methanol and glycerol. Finally, dry air is used to remove water from the product. The simulation is solved for steady-state operation and produces a wide variety of chemical and physical information ranging from throughput to heat duties of individual equipment. It is assumed that crude palm oil can be directly fed into this processing line.

Financial analysis

This scenario involves the following steps: 1. downloading market data from the website of Chicago Mercantile Exchange 33,34 , 2. performing a feasibility analysis based on the downloaded data and physical data provided by Aspen Plus 25 model of a chemical plant converting crude palm oil into biodiesel, see Fig. 5. It is important to note that both futures markets deal with nancially settled contracts, but it is assumed that their prices are an accurate approximation of the physically settled ones. For transportation purposes it is assumed that the delivery location of crude palm oil is in Malaysia and of biodiesel in southern China. The contract details can be seen in Table 1. A plot of the prices can be viewed in Fig. 3. Table 1: Details of the futures contracts. Crude palm oil Exchange Contract Unit Currency Delivery location

Chicago Mercantile Exchange 25 t USD Malaysia

33

Biodiesel

Chicago Mercantile Exchange 34 100 t USD Southern China

The required quantities of crude palm oil contracts, utilities (steam, water, heating fuel and electricity), transport to and from the plant and storage volume are calculated and costed per tonne of biodiesel. The prices were determined based on the plant specications (provided in the documentation of Lurgi GmbH's biodiesel plant design) and the literature

10


Page 10 of 28

WHR-EX

EXHB

B1

OIL

PREHEAT

EXH1

EXH3

HEATED

10E01

MEOH

EXH2

MX01

OIL1

NAMEOH

10D01

EXH4

ESTER 10E02

ESTER1

INTMD1


11 GLYC

ESTER2

10D02D

10D02

ESTER3

10E03

VR1

10D03

ESTER4

10E04

VENT1

ESTER5

VR2

INTMD2

VENT2

10D04

10D03D

MX02

WATER

10E05

MERECYCL

ESTER6

GLYC1

MEFEED

10D07

CONFEED

ME+WATER

MX03

HOTAIR

ESTER7

MIXFEED

10D08

10P01

PFEED

10D06

STILL

FINALPRD

MOISTAIR

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

EXWATER

Page 11 of 28 Industrial & Engineering Chemistry Research

Figure 2: Aspen Plus model of a crude palm oil-to-biodiesel plant (steam generation equipment not shown).


Page 12 of 28

(references provided in Table 2). The summary of prices can be found in Table 2. To calculate the prot the programme solves eqn. 1 outlined in Section . The best contract and storage schedule combination is chosen based on this information using an exhaustive search, results of which can be seen in Fig. 4. The result of the analysis from 11.06.2018 is 219.28 USD per tonne of biodiesel can be made. This can be achieved by buying contracts for delivery of crude palm oil in July 2018 and selling contracts for delivery of biodiesel in August 2018 in a ratio of 4 to 1. The largest cost contribution comes from transportation. Note that it is assumed that all crude palm oil was converted instantaneously (relatively to the timescales involved). Duration of storing the reagent and the product is irrelevant as their storage costs are the same, as visible in Fig. 4. This result indicates that there may be a realistic scope for increasing protability of a chemical plant by exploiting the opportunities across dierent commodity markets. It is also shown that this can be achieved in an automated manner. Table 2: Storage, transport and utility prices (all quantities that do not have a reference are provided in the documentation of Lurgi GmbH's biodiesel plant design.) Storage

Crude palm oil Biodiesel Crude palm oil Biodiesel 35 36

35

Electricity Fuel gas Steam (low pressure) Steam (high pressure) Cooling water Process water

0.2075 0.2075

USD per month-tonne USD per month-tonne

Transport

5.0 40.0

USD per tonne USD per tonne

Utilities

0.0000386 0.0000093 0.01264 0.00349 0.00174 0.00033

12

SGD USD USD USD USD USD


per per per per per per

kJ kJ kg kg kg kg

Page 13 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3: Plot of the crude palm oil and biodiesel futures prices from 11.06.2018 against expiry dates, so-called futures curves. For example, a contract for future delivery of crude palm oil in September 2018 can be bought (or sold) at approx. 605 USD per tonne. Natural gas-to-methanol conversion Methanol plant simulation

A process producing around 56.791 tonnes per hour (1085 mmBTU per hour) of methanol was modelled in Aspen HYSYS V8.8 using the Peng-Robinson property method 37 . The ow sheet model can be divided into two parts: natural gas-to-syngas conversion and syngasto-methanol conversion. Overall, they include two reaction steps (modelled using plug ow reactors - PFRs), three separation steps, three recycle loops, a water top-up section and auxiliary equipment, see Fig. 2. Natural gas (mostly composed of methane, but also ethane and propane) is stripped of any pollutants (such as sulphur compounds) and then converted into syngas (a mixture of carbon monoxide and hydrogen) via steam reforming. Water is removed from the mixture, which is then converted into methanol in a PFR. Lastly, the product is puried up to the desired specications. The simulation is solved for steady-state

13



Table 3: Prices and expiry dates of futures contracts for crude palm oil and biodiesel futures prices from 11.06.2018. Contract expiry (-) Jul-18 Aug-18 Sep-18 Oct-18 Nov-18 Dec-18 Jan-19 Feb-19 Mar-19 Apr-19 May-19 Jun-19 Jul-19 Aug-19 Sep-19 Oct-19 Nov-19 Dec-19

Crude palm oil price (USD per tonne) 592.5 592.75 596.25 601.25 608.5 613.25 614.75 618.25 622 621.5 620.75 621 621.25 620.75 620 619.5 619 618.5

Biodiesel (USD per tonne) 850.5 854.932 854.402 854.088 832.859 831.057 830.013 843.068 842.188 840.417 854.286 853.826 853.75 842.263 842.263 842.263 842.263 842.263

operation and produces a wide variety of chemical and physical information ranging from throughput to heat duties of individual equipment.

Financial analysis

This scenario involves the following steps: 1. downloading market data from the websites of Chicago Mercantile Exchange 23 and Zhengzhou Commodity Exchange 24 (the contract details can be seen in Table 4, while a plot of the prices can be viewed in Fig. 6), 2. performing a feasibility analysis based on the downloaded data and physical data provided by Aspen HYSYS 26 model of a chemical plant converting natural gas into methanol, see Fig. 5. The required quantities of natural gas contracts, utilities (steam, water, heating fuel and 14


Page 14 of 28

Page 15 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 4: Details of the futures contracts. Natural gas Exchange Contract Unit Currency Delivery location

Chicago Mercantile Exchange 10 mmBTU USD Henry Hub pipeline

23

Methanol

Zhengzhou Commodity Exchange 24 1 tonne (approx. 19.1 mmBTU) CNY Zhengzhou Exchange

electricity), transport to and from the plant and storage volume are calculated and costed per tonne of methanol. The prices were determined based on the plant specications (based on the documentation of Lurgi GmbH's biodiesel plant design) and the literature (references provided in Table 5). The summary of prices can be found in Table 5.To calculate the prot the programme solves eqn. 1 outlined in Section . The best contract and storage schedule combination is chosen based on this information using an exhaustive search, results of which can be seen in Fig. 7. The analysis from 11.06.2018 recommends that no trade should be made in order to avoid making a loss. The least loss of approximately 70284 USD per tonne of methanol can be achieved by buying contracts for delivery of natural gas in September 2018 and selling contracts for delivery of methanol in December 2018 in a ratio of 3 to 1. The largest cost contribution comes from the utilities. Note that it is assumed that all natural gas was instantaneously (relatively to the timescales involved) converted on arrival and methanol was stored until delivery date. However, as visible in Fig. 7 the impact of storage cost is relatively small. This result indicates that direct arbitrage with natural gas may not be possible since the markets are ecient (i.e. market prices reect all relevant information) and the the increased cost to be paid in Asia, the so-called Asian Premium reects the additional costs for (i.e. transport and storage). Additionally, it is cheaper to produce methanol at the source of natural gas as the former is signicantly cheaper to transport than the latter.

15



Page 16 of 28

Table 5: Storage, transport and utility prices (all quantities that do not have a reference are based on the documentation of Lurgi GmbH's biodiesel plant design). Natural gas Methanol

Storage

38

0.121 0.2075

Transport

39

Natural gas Methanol 40

36

USD per month-mmBTU USD per month-tonne

approx. 7 22.9

USD per mmBTU USD per tonne

Utilities

Electricity Fuel gas Steam (low pressure) Steam (high pressure) Cooling water Process water

0.0000386 0.0000093 0.01264 0.00349 0.00174 0.00033

SGD USD USD USD USD USD

per per per per per per

kJ kJ kg kg kg kg

Table 6: Prices and expiry dates of futures contracts for natural gas and methanol futures prices from 11.06.2018. Contract expiry (-) Jul-18 Aug-18 Sep-18 Oct-18 Nov-18 Dec-18 Jan-19 Feb-19 Mar-19 Apr-19 May-19

Natural gas price (USD per mmBTU) 2.89 2.894 2.876 2.89 2.937 3.047 3.133 3.102 3.004 2.637 2.595

16


Methanol (CNY per tonne) 2719 2800 2806 2822 2848 2872 2892 2861 2819 2777 2731

Page 17 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4: Plot of marginal prot per tonne of biodiesel against dierent dates of crude palm oil and biodiesel deliveries based on futures prices from 11.06.2018. For example, the rst point corresponds to accepting a delivery of crude palm oil in July 2018 and delivering biodiesel in August 2018. 17 ACS Paragon Plus Environment

301017

301302


18

301303

E-301005

301018

R

301202

301003

301002

VLV-301009

C-301002

VLV-301001

301019

V-301005

RCY-1

301001

301211

C-301002 ENERGY

301304

301004

301020

MIX-301001

301305

E-301006

301021

301005 E-301001

301203

V-301006

301022

301900

V-301001

301006

301310

VLV-301010

C-301003

301218

301212

C-301003 ENERGY

301023

301311

301214 E-301011

301901 301200

301007 V-301002

R

301024

RCY-3

R

MIX-301002

RCY-2

301308 E-301010

301216

301012

301009 E-301002

301025

301213

301309

P-301003 ENERGY

P-301003

301200'

301011

301902

V-301003

301008

A

301217

ADJ-2

301026

301010

301225

301209

301210

301051

301027

MIX-301003

E-301007

301215

E-301003

R-301001 ENERGY

A

TEE-301003

ADJ-1

R-301001

ADJ-3

A

301226

E-301008

R-301002

P-301002 ENERGY

301013

301300

301306

301050

301201

301014

301052

P-301002

301301

E-301004

301307

E-301009

301053

301015

V-301004

301054

301204

301205

301056

TEE-301001

VLV-301002

TEE-301002

C-301001 ENERGY

V-301007

C-301001

301016

301024'

301017

C-301004 ENERGY

301057

C-301004

V-301008

301055

301003'

301206

T-301001

301018

T-301001 REBOILER

301058 T-301001 CONDENSER

301303

E-301005

301207

C-301002

T-301002 REBOILER

301304

301305

301021

P-301001

E-301006

P-301001 ENERGY

301208

301250

301020

301211

C-301002 ENERGY

T-301002 CONDENSER

VLV-301009

T-301002

301202

V-301005

301019

301203

V-30100

301022

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 301302

Industrial & Engineering Chemistry Research Page 18 of 28

Figure 5: Aspen HYSYS model of a natural gas-to-methanol plant. The upper section converts natural gas into syngas, while the lower syngas into methanol.

Page 19 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6: Plot of the natural gas and methanol futures prices from 11.06.2018 against expiry dates. For example, a contract for future delivery of methanol in September 2018 can be bought (or sold) at approx. 9.9 USD per mmBTU.

19



Figure 7: Plot of marginal prot per tonne of methanol against dierent dates of natural gas and methanol deliveries based on futures prices from 11.06.2018. For example, the rst point corresponds to accepting a delivery of natural gas in July 2018 and delivering methanol in August 2018. 20


Page 20 of 28

Page 21 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Conclusions and future work This paper presents an implementation of an automated price-discrepancy spotter powered by market and physical data which has the ability to support production planning, nancial risk governance and supply chain management thereby enhance overall economic productivity. This ability is illustrated for two scenarios: conversion of natural gas to methanol and crude palm oil to biodiesel which serve as case studies. The programme searches for opportunities to make additional prot by analysing the futures curves, i.e. the market prices anticipated by the futures market, for both the reagent and the product. It considers cost of storage and conversion (other feedstock, steam, electricity and other utilities) derived from physical simulations of the chemical process. The two case studies considered are natural gas-to-methanol conversion and crude palm oil-to-biodiesel (FAME) conversion. In case of the former, an economic loss can be avoided by foregoing procurement and conversion of natural gas to methanol. In case of the latter, on the contrary, 219.28 USD per tonne of biodiesel can be earned by buying contracts for delivery of crude palm oil in July 2018 and selling contracts for delivery of biodiesel in August 2018 with an increased margin of 11.6%. Based on the two analyses presented it can be concluded that there is realistic scope for increasing protability of a chemical plant by exploiting the opportunities across dierent commodity markets and over time by planning procurement, production and storage in an automated manner utilising articial intelligence. Additionally, the results suggest that direct arbitrage with natural gas may not be possible as the markets are ecient and transporting natural gas tends to be more expensive than methanol. Hence, the so-called Asian premium, the extra amount charged for fossil resources in Asia, is purely owing to transportation cost. This study therefore exemplies that cyber-physical systems can eectively combine comprehensive engineering models with real-time nancial data to optmise the output and economics of a chemical plant. Future work will involve employing the automated price-discrepancy spotter in conjunction with J-Park Simulator (JPS) 1,3 . The JPS is a cyber-physical system for designing, 21



computer-aided process engineering (CAPE) and managing an eco-industrial park (EIP). This combination will enable the application of the current ndings to larger networks and dierent types of commodities, machine-to-machine market creation (spot and futures) and introduction of more complex trade deals. Additionally, a greater variety of models and dynamic behaviour (e.g. a simulation of a process line during start-up and shut-down) can be introduced.

Acknowledgement This project is funded by the National Research Foundation (NRF), Prime Minister's Oce, Singapore under its Campus for Research Excellence and Technological Enterprise (CREATE) programme. The authors would like to thank Churchill College, Cambridge for their continual support.

References (1) Kleinelanghorst, M. J. et al. J-Park Simulator: Roadmap to Smart Eco-Industrial Parks; 2016; Date accessed: 09.12.2016. (2) Kraft, M.; Mosbach, S. The future of computational modelling in reaction engineering. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences

2010,

368, 36333644.

(3) Pan, M.; Sikorski, J.; Kastner, C. A.; Akroyd, J.; Mosbach, S.; Lau, R.; Kraft, M. Applying Industry 4.0 to the Jurong Island Eco-industrial Park. Energy Procedia 2015, 150. (4) Pan, M.; Sikorski, J.; Akroyd, J.; Mosbach, S.; Lau, R.; Kraft, M. Design technologies for eco-industrial parks: From unit operations to processes, plants and industrial networks. Applied Energy

2016,

175, 305 323. 22


Page 22 of 28

Page 23 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(5) Kastner, C. A.; Lau, R.; Kraft, M. Quantitative tools for cultivating symbiosis in industrial parks; a literature review. Applied Energy

2015,

155, 599 612.

(6) Garcia, P.; Leuthold, R. M. A selected review of agricultural commodity futures and options markets. European Review of Agriculture Economics

2004,

31, 235272.

(7) Szakmary, A. C.; Shen, Q.; Sharma, S. C. Trend-following trading strategies in commodity futures: A re-examination. Journal of Banking & Finance

2010,

34, 409426.

(8) Fung, W.; Hsieh, D. A. The Risk in Hedge Fund Strategies: Theory and Evidence from Trend Followers. Review of Financial Studies

2001,

14, 313341.

(9) Campa, J. M. Multinational Investment under Uncertainty in the Chemical Processing Industries. Journal of International Business Studies

1994,

25, 557578.

(10) Smith, C. W.; Stulz, R. M. The Determinants of Firms' Hedging Policies. The Journal of Financial and Quantitative Analysis

1985,

20, 391.

(11) Bjorgan, R.; Chen-Ching Liu,; Lawarree, J. Financial risk management in a competitive electricity market. IEEE Transactions on Power Systems

1999,

14, 12851291.

(12) Tanlapco, E.; Lawarree, J.; Liu, C. C. Hedging with Futures Contracts in a Deregulated Electricity Industry. IEEE Power Engineering Review

2002,

22, 5454.

(13) Spinler, S.; Huchzermeier, A.; Kleindorfer, P. Risk hedging via options contracts for physical delivery. OR Spectrum

2003,

25, 379395.

(14) Ding, Q.; Dong, L.; Kouvelis, P. On the Integration of Production and Financial Hedging Decisions in Global Markets. Operations Research

2007,

55, 470489.

(15) Kannegiesser, M.; Günther, H.-O.; van Beek, P.; Grunow, M.; Habla, C. Value chain management for commodities: a case study from the chemical industry. OR Spectrum 2009,

31, 6393. 23



Page 24 of 28

(16) Ryu, J.-H. Multiperiod Planning Strategies with Simultaneous Consideration of Demand Fluctuations and Capacity Expansion. Industrial & Engineering Chemistry Research

2006,

45, 66226625.

(17) Park, J.; Park, S.; Yun, C.; Kim, Y. Integrated Model for Financial Risk Management in Renery Planning. Industrial & Engineering Chemistry Research 2010, 49, 374380. (18) Ji, X.; Huang, S.; Grossmann, I. E. Integrated Operational and Financial Hedging for Risk Management in Crude Oil Procurement. Industrial & Engineering Chemistry Research

2015,

54, 91919201.

(19) Longinidis, P.; Georgiadis, M. C.; Kozanidis, G. Integrating Operational Hedging of Exchange Rate Risk in the Optimal Design of Global Supply Chain Networks. Industrial & Engineering Chemistry Research

2015,

54, 63116325.

(20) Kwon, H.; Tak, K.; Cho, J. H.; Kim, J.; Moon, I. Integrated Decision Support Model for Hedge Trading and Production Planning in the Petrochemical Industry. Industrial & Engineering Chemistry Research

2017,

56, 12671277.

(21) Guo, X.; Jiang, X.; Wong, W.-K. Stochastic Dominance and Omega Ratio: Measures to Examine Market Eciency, Arbitrage Opportunity, and Anomaly. Economies 2017, 5, 38. (22) Ng, P.; Wong, W.-K.; Xiao, Z. Stochastic dominance via quantile regression with applications to investigate arbitrage opportunity and market eciency. European Journal of Operational Research

2017,

261, 666 678.

(23) Group, C. Henry Hub Natural Gas Futures Quotes. 2017; http://www.cmegroup.com/

trading/energy/natural-gas/natural-gas.html, Date accessed: 29.05.2017. (24) Exchange, Z. C. Methanol contract daily data. 2017;

24


http://english.czce.

Page 25 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


com.cn/enportal/DFSStaticFiles/Future/EnglishFutureQuotesMA.htm, Date accessed: 29.05.2017. (25) AspenTech, aspentech - Aspen Plus V8.8. 2017; http://www.aspentech.com/

products/Aspen-Plus/V88/, Date accessed: 29.05.2017. (26) AspenTech, aspentech - Aspen HYSYS V8.8. 2017; http://www.aspentech.com/

products/aspen-hysys/, Date accessed: 29.05.2017. (27) Foundation, P. S. Python 3.5.0. 2017; https://www.python.org/downloads/release/

python-350/, Date accessed: 29.05.2017. (28) Oanda, Currency converter. 2017; https://www.oanda.com/currency/converter/, Date accessed: 29.05.2017. (29) XE.com, Currency converter. 2017; http://www.xe.com/currencyconverter/, Date accessed: 29.05.2017. (30) Coinnews Media Group LLC, US Ination Calculator. 2017;

http://www.

usinflationcalculator.com/, Date accessed: 29.05.2017. (31) Microsoft, COM Technical Overview. 2015; https://msdn.microsoft.com/en-us/

windows/desktop, Date accessed: 02.06.2015. (32) Gmehling, J.; Wittig, R.; Lohmann, J.; Joh, R. A Modied UNIFAC (Dortmund) Model. 4. Revision and Extension. Industrial & Engineering Chemistry Research 2002, 41, 16781688. (33) Group, C. USD Malaysian Crude Palm Oil Calendar Futures Quotes. 2017;

http://www.cmegroup.com/trading/agricultural/grain-and-oilseed/ usd-malaysian-crude-palm-oil-calendar.html?optionProductId=8075, accessed: 29.05.2017.

25


Date


Page 26 of 28

(34) CME Group, FAME 0 Biodiesel FOB Rdam (Argus) (RED Compliant) Futures Quotes.

http://www.cmegroup.com/trading/energy/refined-products/

2017;

fame-0-argus-biodiesel-fob-rdam-red-compliant-swap-futures.html,

Date

accessed: 29.05.2017. (35) Fadhil, M. SE Asia-to-China palm oil freight rates may fall on weak demand.

https://www.icis.com/resources/news/2013/11/08/9723077/

2013;

se-asia-to-china-palm-oil-freight-rates-may-fall-on-weak-demand/,

Date

accessed: 29.05.2017. (36) Energy Market Authority, Electricity Taris. 2017; https://www.ema.gov.sg/Non_

Residential_Programmes_Electricity_Tariffs.aspx, Date accessed: 29.05.2017. (37) Peng, D.-Y.; Robinson, D. B. A New Two-Constant Equation of State. Industrial & Engineering Chemistry Fundamentals

1976,

15, 5964.

(38) Comission, F. E. R. Current state of and issues concerning underground natural

gas

storage.

2004;

https://www.ferc.gov/EventCalendar/Files/

20041020081349-final-gs-report.pdf, Date accessed: 29.05.2017. (39) EY, Competing for LNG demand:

the pricing structure debate. 2014; https:

//webforms.ey.com/Publication/vwLUAssets/Competing-for-LNG-demand/ $FILE/Competing-for-LNG-demand-pricing-structure-debate.pdf,

Date

ac-

cessed: 29.05.2017. (40) methanex, Methanex Investor Presentation. 2016; ttps://www.methanex.com/sites/

default/files/investor/MEOH%20Presentation%20-%20June%202016_0.pdf, Date accessed: 29.05.2017.

26


Page 27 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Graphical TOC Entry

27


Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34


Page 28 of 28

Enhanced Procurement and Production Strategies ... - ACS Publications

Recommend Documents