USING INTERRUPTIBLE TRANSPORTATION SERVICES
This paper discusses whether introducing interruptible transportation services in a natural gas network can increase throughput without deteriorating the security of supply. In our analysis we include both firm and interruptible transportation services, where firm services are characterized by a guaranteed level of security of supply while interruptible services are delivered provided there is available capacity on the given day. We develop a general model framework for analysis of interruptible transportation services, and present results from a case study based on realistic data from the Norwegian natural gas transportation system that currently covers nearly 20% of European gas consumption. We have also used a stylized representation of the booking regime at the NCS for our analysis.
Interruptible transportation services are well known within the natural gas supply chain, as they are available in the US and in several European systems (including the Norwegian). These services allow the Transportation System Operator (TSO) to oversell the capacity by reselling capacity that is booked firm but not nominated, without relieving the obligation to the original buyer. It is usually required that all firm capacity, defined by a predefined static limit, is sold before any capacity can be resold as interruptible. The intention of the interruptible services is to improve the short term redistribution of transportation capacity to support an efficient use of the network. Our motivation for introducing interruptible transportation services is different. We focus on increasing the capacity initially made available by the TSO rather than redistribution of allocated capacity between the producers. The latter will increase the utilization of the offered capacity in the network, while the former will increase the capacity offered.
A high security of supply level is important on the market side, for the shippers to be able to deliver in long-term contracts. It is also important on the production side, to ensure that the oil production on the fields with associated gas will not be impeded. In order to maintain a high level of security of supply on firm services, it is necessary for the system operator to withhold some capacity in the system at the time of booking to have flexibility to handle uncertainties in the final operation. This withheld flexibility can decrease the capacity utilization in the network. Security of supply can be expressed through different measures. We define the security of supply level as the expected throughput in the whole system relative to the total firm booking. Unplanned events, such as outages and technical failure, cause uncertainty in the available capacity in the transportation network. Furthermore, the system operator must take into account system effects that make it impossible to a priori determine static capacities.
The short-term system flexibility in our analysis comes from the possibility to increase production levels in some fields and to reroute the gas. Linepack is not included in our single-period model. The availability of linepack would have increased the network flexibility that could reduce the value of introducing interruptible services, while in a realistic dynamic setting this effect would highly depend on the time structure of network events and the trade-off between commercial use and security of supply considerations.
The objective of this case study is to evaluate the potential from introducing interruptible transportation services for the network as a whole, while recognizing that different agents in the system have different incentives. We include two decision makers in the network: the transmission system operator (TSO) and a shipper of natural gas. The TSO is responsible for the routing of gas in the network and allocates capacity to the shipper to ensure that the security of supply in the network is within given bounds. The TSO can offer two different types of transportation services: firm and interruptible. Only firm services have a security of supply measure, while the interruptible services can freely be interrupted by the TSO whenever the available capacity in the transportation network is not sufficiently large.
Our modelling framework consists of five optimization models that are run in sequence. This approach gives a stylized representation of the decision making process at the Norwegian Continental Shelf (NCS). It is important to note that the analysis does not cover all the dynamics in the current regime, and as such our results will not be directly applicable to the NCS-system, but they will indicate the potential gains from introducing interruptible contracts in general transportation networks where security of supply is important. In the first shipper problem, the producer requests booking based on possible event and market price outcomes and estimates of future interruptions, production and sales decisions. Based on the booking request the TSO allocates firm capacity seeking to maximize the correspondence with the booking request. In the second shipper problem interruptible booking decisions are taken. When all booking is decided and the events in the transportation network have become known the TSO decides how much interruptible and firm capacity he needs to interrupt. Based on the final available capacity and the realized market prices the shipper decides on the amounts to produce and sell.
Production cost functions
The shipper’s production cost function consists of two parts, one for production with associated oil, called `must-take’, and one for the swing field production. The reasoning behind this is that each field node in our network represents a mix of must-take and swing fields. The must-take production is closely linked to the oil production, so if the gas production is decreased, the oil production must also be decreased. Such a decrease will lead to a substantial loss for the shipper. For the swing fields, the gas production will not influence the oil production.
Gas network topology
The gas network in our case study is based on the infrastructure on the NCS, and is illustrated in Figure 1. Our network has a maximum delivery capacity of 351 MSm3, while the largest daily delivery from NCS in 2011 was 361 MSm3. The eight fields that we use in our case study represent approximately 50 real fields, aggregated by region. These aggregated fields cover both must-take fields and swing fields, and the swing fields imply a larger daily production capacity than transportation capacity. All fields and markets are booking nodes in the network, such that they require booking of transportation capacity corresponding to their production and sale, respectively. The booking tariffs for firm transportation capacity correspond to the real tariffs on the NCS, defined individually for each booking node. We have assumed the booking tariff for interruptible services to be half the price of firm services. The model has been tested with security of supply requirements for firm services in the range from 0.99 to 1 (where 1 indicates that all firm capacity must be delivered in all scenarios).
Figure 1: The network used for our case study. This is an aggregated representation of the network at the NCS.
We have generated price outcomes based on real spot prices from 2010 and 2011 for all market hubs directly connected to the NCS export network. The market prices are represented by 10 outcomes that are generated with a procedure matching the statistical moments and correlations in the observed prices. Testing our model framework on multiple sets of generated market price outcomes showed that 10 was a sufficient number of outcomes to achieve stable results. The unplanned events in the network are described by artificial data. We have defined 19 events with reduced capacity, each corresponding to a separate outcome. In addition, we have a default outcome where the system operates at full capacity. The probability and extent of the capacity reductions are calibrated so that the availability corresponds to the average availability figures reported by Gassco in 2010 and 2011. In total the 10 market outcomes and 20 event outcomes give 200 scenarios.
To discuss our results and quantify the effect of introducing interruptible contracts, we have also used a benchmark which is calculated with the same model framework. In our benchmark, there are no interruptible services available. In the following we use the label `Without’ for the benchmark solution, while `With’ indicates tests where interruptible services are included. We test the two model setups for increasing security of supply requirements, and compare the effects on booking levels, total throughputs, incomes and costs.
Our first observation is that the total booking stays constant independent of the security of supply level when interruptible services are available. This can be seen in Figure 2 where the firm booking decreases, while the interruptible booking increases with the same amount. This observation comes from the fact that tariffs are below the marginal value of capacity so that it is the shipper’s view on network capacity that limits the booking. Since the shipper’s preference for transportation capacity is not reduced when booking requests are not fully fulfilled, he will seek to obtain the same total amount of capacity by increasing the interruptible booking. We have also performed a sensitivity analysis where the interruptible tariff is increased to 50% above the firm tariff. The results from this sensitivity analysis correspond to the original analysis.
Figure 2: Resulting booking levels in the analysis.
Our second observation is that the shipper finds the flexibility of unbalanced booking valuable. When allowed, the shipper consistently books nearly 90 MSm3 more entry capacity than exit capacity, even though it implies paying for some transportation capacity that necessarily will be interrupted. This value comes from the ability to adapt to events by substituting production with fields that are not affected by an event. Since the tariffs in the network are very small, less than 13% of the average spot price, the option cost of this flexibility is very low.
Throughput in the system
When the security of supply level is increased, the allocated firm capacity is decreased. This is an expected result since increasing buffers are needed to withstand the events in the network. Our third observation is that the benchmark has a falling expected throughput as the security of supply requirement increases (see illustration in Figure 3). On the other hand, according to our fourth observation, the expected throughput is insensitive to the security of supply requirement when interruptible services are available. These two last observations together confirms our hypothesis, that including interruptible services to the transportation service regime can increase the efficiency by enabling a larger expected throughput in the transportation network without reducing the security of supply. The expected throughput increases with 24% when introducing interruptible services at the lowest security of supply level (0.99), and the difference increases to over 269% when the security of supply requirement is 1.
Figure 3: Total throughput in the system with and without interruptible contracts.
Income and costs
Resulting income shows a pattern similar to the total flow, with an increase of between 24% (with security of supply level of 0.99) to 267% (with security of supply level of 1) compared to the benchmark. Differences in average achieved spot prices are small, which is reasonable since both booking and interruption are allocated before spot prices become known. The ability to adapt to the varying spot prices is therefore limited. Our fifth observation on the other hand, is a dramatic increase in production cost due to decreased oil production for the benchmark as security of supply increases. When the transportation capacity is reduced, natural gas production must be reduced and therefore also the oil production is reduced. This effect is largest in the benchmark since the transportation capacity falls below the must-take production capacity in some fields when the security of supply requirements is high. Since we do not have real production cost functions available there is substantial uncertainty with respect to the true monetary cost of this decreased oil production. The profit margins of oil is however substantially larger than for gas, so the shape of the production cost functions are representative. We therefore also argue that the improved ability of stable oil production through introduction of interruptible services is valid.
To test the significance of the agents’ differing incentives on the supply chain performance, we ran our case study with an alternative set of TSO models where the incentives were more in line with the shipper’s incentives. That is, the original objective functions were replaced with an objective where social surplus in the network were maximized. This corresponds to an idealized situation where the TSO have all price and production cost information. The sixth observation is that the profits can increase if the TSO maximizes social surplus rather than taking market signals from the booking requests only. Due to increased flow and income there is approximately a 10 % increase of profit for security of supply requirements less than 0.997 when interruptible services are available. For stricter security of supply requirements the profit increases are less regular. With interruptible services the model where the TSO maximizes social surplus avoids withholding must-take production with valuable associated oil, which causes a major profit increase of 69 % when security of supply is 1.
Based on our analysis of a test case with a network resembling the network and situation at the Norwegian Continental Shelf there seems to be a potential value in introducing interruptible transportation services to allow for a decreased security margin. Both total flow and income in the system is increased compared to the benchmark solution where interruptible services are not available. It should be noted that the comparison with the current situation at the NCS is simplified and based on our model representation. We have represented the formal regime to the best of our ability, but the current regime also allows for a more dynamic handling of events that are not possible to capture with the analysis framework. As such, our analysis illustrates that the type of contracts we have introduced in this paper can bring value to a general network where security of supply is important, but the numbers do not directly reflect the potential efficiency gains in the current regime at the NCS.
By Marte Fostad, Kjetil Trovik, Midthun, Asgeir Tomasgard
Marte Fodstad holds a PhD in Industrial Economics from the Norwegian University of Science and Technology. She has for more than 10 years been a research scientist at the research institute SINTEF. Her main area of research has been operations research applied within the natural gas and hydro power industries.
KJETIL TROVIK MIDTHUN
Kjetil Trovik Midthun works as Research Manager at the research institute SINTEF. He holds a PhD in Business Economics from the Norwegian University of Science and Technology. His main areas of research have been economic analysis, planning and handling of uncertainty within the natural gas value chain.
Asgeir Tomasgard is Professor in managerial economics and operations research at the Department of Industrial Economics and Technology Management at NTNU. He is also the Director of the Center for Sustainable Energy Studies (CenSES), and holds a part time position at SINTEF as a Senior Researcher.
 SINTEF Energy Research, Energy Systems, Postbox 4761 Sluppen, 7465 Trondheim, Norway  SINTEF Technology and Society, Applied Economics, Postboks 4760 Sluppen, 7465 Trondheim, Norway  Norwegian University of Science and Technology, Dept. of Industrial Economics and Technology Management, Sentralbygg I, Alfred Getz veg 3, 7491 Trondheim, Norway  Fodstad, M., Midthun, K.T. and Tomasgard, A., “ Adding flexibility in a natural gas transportation network using interruptible transportation services”, European Journal ofOperational Research, Volume 243, Issue 2, 2015, Pages 647-657