Experimental Market Design and Methodology: The Alaska Gas Pipeline

\(\newcommand \ensuremath [1]{#1}\) \(\newcommand \footnote [2][]{\text {( Footnote #1 )}}\) \(\newcommand \footnotemark [1][]{\text {( Footnote #1 )}}\)

2 Background

Oil and gas resource development commonly follows a time-line of exploration, discovery, and pipeline and process facilities construction, followed by delivery to markets over the economic lifetime of resource availability.

The inherent geological, technological and market uncertainty over long time horizons is further complicated by the sequential supply chain form of that development, and its potential for creating bottlenecks.

Moreover, the interests, fortunes and composition of the pipeline users are subject to constant change. The result is baffling complexity and frustration for government regulators charged with the responsibility to maintain open access for other than incumbent players without interfering with incentives for efficient economic development.

The rich technological and economic issues are illustrated by the Prudhoe Bay field in Arctic Alaska. Discovered in 1969, Prudhoe became the largest reservoir in North America, and, through enhanced recovery technologies, is expected to yield over 13 billion barrels of oil; this is some 35 percent more than initially estimated based on the original hypothesized technology so much has that technology been improved. In 2017Prudhoe production was down to 7 percent of total US output, but its output has been declining since 1988 after yielding more than 1.5 million barrels per day for a decade.

The Alyeska Pipeline (Operator of TAPS-Trans Alaska Pipeline System), planned in 1969, approved in 1973, and completed in 1977, accommodates the transportation of crude oil to remote markets. The jointly produced gas has no external market, and is mostly recycled to maintain lift head in the reservoir as articulated below. The accumulating gas reserves in the oil reservoirs, and the many gas discovery wells, capped until and if the gas can be shipped to market, has led to a proposed gas pipeline. We describe below the (1) oil and gas recovery and transportation technologies, (2) proposed market models for transferring shares in the joint ownership, (3) operation of the various technological modules in the supply chains, and (4) proposed research and its time-line of execution.

2.1 Engineering economics of oil and gas, dynamics of oil supply chain development

The Prudhoe discovery launched an ongoing dynamic economic response that resulted in the construction of a supply chain connecting North Slope exploration activity, wellheads, processing facilities, and the pipeline transportation of crude oil to the tanker port of Valdez 800 miles to the south.

By 2001 the original Prudhoe strike had been joined by 17 smaller new fields beginning in 1969 with the discovery of Kuparuk, the second largest field in North America. A map showing these fields and associated processing and transportation system is shown in Figure ??.

Figure 2: Map of North Slope reservoirs and facilities, reproduced from NorthSlope2004.

fig:gasfieldmap

The Prudhoe reservoir consists of an oil and gas-saturated mixture, a free gas cap on top of the oil structure and water below the gas-oil mixture. The recovered mixture of oil, gas and water is gathered from multiple wellheads at drill sites, and processed through facilities that separate the gas, water and oil. The oil is cooled to enter the first pipeline pump station for shipment to Valdez. The water, along with additional processed seawater to replace the oil, is re-injected as flood-water back into the reservoir to help maintain the lift pressure that sweeps oil from the pores of the reservoir rock. Some natural gas liquids are extracted by refrigeration from the gas, and shipped on the pipeline with the crude oil. Some gas and liquids are re-injected to mix with the reservoir oil to enhance recovery, while some gas is routed to a Central Gas Facility, compressed and re-injected into the gas cap to maintain pressure.

Figure 3: Engineering flow model of the processing facilities, reproduced from NorthSlope2004.

Figure 3 provides a generic engineering flow model of the processing facilities used to separate the oil, gas and water mixture entering from the wellheads. The oil, gas and water mixture comes in from the wellheads on the lower left of Figure 3. Gas, miscible with oil, is recycled into the reservoir through injection wells. Natural gas liquids are transported to the pipeline (TAPS) along with the oil. Dry gas is compressed and injected into the reservoir gas cap. The second and third separation steps remove the water, which is injected back into the field.

As older reservoir fields mature and new fields are discovered it alters the facility and pipeline needs of new and incumbent users. For mature fields the recovered oil content declines relative to recovered water and gas. This alters the facility and pipeline capacity and operating needs of the original owners to the extent that they are inactive or unsuccessful in new exploration. As new fields are discovered and developed, the mix of owners may be distinct from the original incumbents, there is a demand for entry access to capacity and operations.

In petroleum reservoirs, the processing needs are for oil, gas and water, any one of which may be capacity constrained in a given process facility, and impact pipeline access. Some fields like Kuparuk are too far from Prudhoe to share processing facilities, operate their own separation facilities, re-inject gas and water, and must pipe the oil across to the TAPS pump station for transport south. Satellite fields located above or below the Prudhoe or Kuparuk formations share their respective processing facilities. Figure ?? shows these fields, some overlapping, some disjoint on the North Slope surface.

In gas reservoirs, gas, water and carbon dioxide must be separated; any component may be capacity constrained in a given process facility; and similarly for pipeline access. Thus, over time, access needs, in the form of capacity rights to the pipeline, to feeder lines, to oil, water, carbon dioxide, and gas processing capacity, will change. It changes because of a changing mix in exploration effort and success, changes in the maturity state of old versus new fields, and changes in technology.

If new wells are brought in, older less productive wells must be either backed out, or the constraints on separating capacity increased. Without a gas pipeline the market for gas is severely limited and beyond local energy needs—heat and power generation—gas value on the North Slope is determined primarily by the increased recovery of the future oil that is derived from the recycled gas.

2.2 The institutional structure

It is not necessary here to delve deeply into the common carrier pipeline regulation structure as our objective is to study various modifications in the joint venture agreements that might be undertaken by the government and developed as a more flexible and efficient alternative to the attempt to use adversarial cost-based regulation to maintain open access.

The [REWRITE] idea we research is for the state to adopt a new role. To define, and enforce pre-specified sharing rights rules that would allow markets to govern the actions of each joint venture cotenant in the supply chain, and assist in scheduling auctions as needed for transferring ownership, leased or rented capacity rights among the cotenants and new entrants in each venture and across ventures.

Entry, exit and access prices emerge from these auctions, not from regulation based on historical cost. Each cotenant and therefore the collective process, is entirely forward-looking and disciplined by opportunity costs. None of the Alaskan incumbent energy firms, or the potential new entrant/investors in a gas pipeline transport system, has monopoly power; each competes in world markets at the delivery end if the facility.

Regulators legitimately desire to achieve open access, under fair rules and want to avoid having some entity ensconced in a dominant bargaining position. These objectives are highly desirable, but it is problematic to achieve them through any historical-cost based price regulation, which, has offered hopes that have been hard to realize in a manner that is satisfactory to all parties.

Pitfalls of cost-based regulation

All cost-based regulation is flawed because it attempts to derive access prices from historical costs in economic environments that are dynamic. At each point in time, efficient facility utilization and investment must always reflect current and expected future revenue and costs, and therefore historical cost is irrelevant. The theory of economic regulation derives from static analysis and has no serious applicability to the oil and gas industry, which experiences highly volatile world prices determined by changing world supplies and demand. Hence, if oil prices fall relative to the expectations that led to the construction of existing facilities, those incumbent facilities suffer capital losses. Under such an eventuality, there is no assurance, only hope, that their historical cost will be recovered; all such costs are irrelevant in determining the price of capacity access. Independents will shun entry, unless they can obtain capacity access at prices that reflect the reduced value of the oil, while incumbents will hope to use the regulatory cost-based rules to recover their historical cost and avoid the capital loss. However, in such a scenario, the willingness-to-pay demand for access is below its historical cost. No economic terms enable the transfer of rights at historical cost.

In contrast, if oil prices rise, historical cost is irrelevant to the prices of capacity access because those prices must reflect the capital appreciation earned by incumbents on their risky up-front investment. The shoe is now on the other foot. Independents will be eager to enter, perhaps expecting a fair regulatory apparatus to deliver them access at historical cost. Incumbents having seen their investments turn into a positive return cannot be expected to be enthusiastic supporters of proposals to share those capital gains with newcomers, who bore not any risk, in the form of access at historical cost. For them it is heads I lose my investment, and have no one to share the losses with; tails I win but must share the gain with late-comers. In that situation, everyone aspires to be a late-comer and take advantage of after-the-fact knowledge of whether the investment will be a winner or a loser.

In either of these scenarios, we have an economic standoff, and the regulatory compact is incompatible with the disjoint interests of the parties. Whatever happens to the price of oil; whatever happens in new exploration; whatever happens to improve the technology of discovery, product recovery, and processing, the relevant price of access to resources must be based on the willingness to buy and willingness to sell by entities whose options and opportunities are known only to those entities, albeit with great uncertainty.

Features of private joint venture contracts

Joint venture facility sharing has common elements across many industries. The facilities might be power lines, power plants, oil and gas reservoirs, pipelines, newspaper printing facilities, and so on, now or in the past. Ownership and operations constitute the two sets of contracting issues.

Joint venture facility sharing is typically based on two agreement documents : ownership wherein each of the co-tenants commonly share capacity in the same proportion and that they share the capital investment by each. If you pay for half the capital costs, then you have drawing rights on half of the pooled capacity.

Operations, wherein each of the cotenants agree to share fixed and variable operating cost depending on capacity used in real time. Often costs are shared in proportion to output, but this is not efficient if output proportions are unequal and variable unit costs are not constant. This is the serial cost sharing problem, addressed in the economic literature for single output facilities, like a pipeline, but unsuitable, without further research, for application to multiple interdependent output facilities like separation plants for oil, gas and water (hmss1992; hm1994; hm1996; eFhM1999a; freidman02; vKolpin1996).

Three standard rules or practices, that have emerged historically in these private contracting arrangements, are the following :

For example, the Arizona Public Service Company (APS) Operates and owns 29.1% of the plant. Its other major owners include the Salt River Project (17.5%) and the El Paso Electric Company (15.8%), Southern California Edison (15.8%), PNM Resources (10.2%), the Southern California Public Power Authority (5.9%) and the Los Angeles Department of Water and Power (5.7%).

These rules operate to limit contestability, block access, and reduce the ability of facility use and expansion to be disciplined by opportunity cost, and therefore are unsuitable for creating an institution that would serve a competitive, self-regulating function in the dynamic energy market.

Competitive property right rules for joint ventures

How can the state modify these rules, as a condition for development and in return forego cost-based regulation, which imposes regulatory adversarial costs on all parties? Here are the property rights rules that we propose to study.

Perhaps operations and its management should be required to be a separate entity shared by the co-tenants in the venture. However, our studies do not include examination of this question.

2.3 Pipeline flow and capacity engineering economics

The following equations come from Smith1962 which is a summary of Chenery1949. To determine capacity \(Q\), we construct the equation for the flow of gas in a pipeline :

\begin{equation} \ln Y = a + (8/3) \ln D + \ln (p_1^2 - b) \label {eq:1} \end{equation}

where :

\(Y\)

is gas flow in the pipe (e.g. cubic feet per minute),

\(D\)

the diameter of the pipe,

\(p_1\)

is inlet driving gas pressure from the source,

\(b\)

is square of delivery outlet pressure \(p_{2}\) which we take as a given constant (or it could be the delivery pressure to the next pump station in a long pipeline).

Inlet pressure cannot exceed the rupture limit of the pipe; i.e., \(p_{1}< \frac {2ST}{D}\) the packing capacity of line; \(T\) is the pipe thickness; \(S\) is the maximum working stress of the pipe material. It follows that throughput capacity \(Q\) (i.e., max \(Y\)) is given by :

\begin{equation} \ln Q = a + (8/3) \ln D + \ln \Biggl ( {\biggl ( \frac {2ST}{D} \biggr )}^2 - b \Biggr ) \label {eq:2} \end{equation}

Now consider the initial pumping capacity decision, \(P\), and how \(Q\) and \(P\) are related. Let \(Y_0\) be the initial designed gas flow in pipe, given \((T, D)\); then you need installed initial pumping capacity, \(P < \frac {2ST}{D}\) given by :

\begin{equation} \ln Y_0 = a + (8/3) \ln D + \ln (P^{2} - b) \label {eq:3} \end{equation}

Initial total construction cost, for a line of given length, will be some variation on an equation of the form :

\begin{equation} TC = K + C (D, S, T). \label {eq:P8} \end{equation}

All the above ignores pump spacing and other details in a real application.

We can write total capacity cost as \(K + C(Q)\), so that when we write \(C(Q)\) it is actually defined parametrically as \(C(D, S, T)\) where each point \((D, S, T)\) maps into corresponding values of \(Q\) and of \(C\).

Contingent on \((D, S, T)\), then \(Y_0\) is determined by \(P\) in equation 2.2. Equations 2.1 and 2.2 tell you how the determination of \(P\) depends on \(Q\).

Next, the idea is to get the various modules laid out for an initial experimental design. Our experimental design will specify a few point capacities such as \(Q1\) and \(Q2\), and their corresponding costs \(C1\) and \(C2\). Our main interest will be to study bidding behavior and the bidding mechanisms for joint sharing of \(Q\).

Individual firm optimization

Let

\(Q\)

be capacity and

\(K\)

be avoidable fixed capital cost of \(Q\).

The total cost of capacity (\(tc\)) is :

\begin{equation} tc = \begin {cases} K + C(Q), & \mathrm {if}\quad Q > 0 \\ 0, & \mathrm {if}\quad Q = 0 \end {cases} \end{equation}

where \(C(0) = 0\), \(C’(Q) > 0\) and \(C”(Q) < 0\), for \(Q>0\).

This implies that \(K\) is an excludable public cost for the members of the Joint Venture.

Let \(i\) denote an active (profitable) members of the joint venture, \(i = 1, 2, \ldots , N\).

Each \(i\) chooses \(Q_i\) to maximize :

\begin{equation} z_i(q_i) = V_i(q_i) - s_i(q_i, q) + t_i(q_i, q), \end{equation}

where :

\(z_i(q_i)\)

is the net profit of firm \(i\),

\(q_i\)

is the share of capacity that is allocated to firm \(i\),

\(V_i(q_i)\)

is the net profit (value) to firm \(i\) of producing \(q_i\),

\(s_i(q_i, q)\)

is the share of the total capital cost paid by firm \(i\),

\(t_i(q_i, q)\)

is the “tax or subsidy” paid by or to firm \(i\) for investing \(k_i\),

\(Q\)

\(= \sum _i q_i\),

\(q_i\)

operating capacity \(i\) + reserve capacity \(i\),

\(k_i\)

share of \(K\) paid by \(i\),

\(K\)

\(= \sum _i k_i\).

Set \(s_i(q_i) = \frac {q_i}{q} \left (\, K + C(Q) \,\right )\); that is, the proportional sharing rule. Since

\begin{align} \frac {d( \frac {u}{v} )}{dx} & = & \frac {v \frac {du}{dx} - u \frac {dv}{dx}} {v^2} \\ \frac {d( \frac {q_i}{q} )}{dq_i} & = & \frac {q \cdot 1 - q_i \cdot 1} {q^2}, \end{align}

then

\begin{equation} ds_i = \frac {q_i}{q^2} \cdot k + \frac {q_i}{q^2} \cdot c(q) + \frac {q_i}{q} \cdot c’(q) ds_i > 0 \end{equation}

is the equation for individual firm optimization.

2.4 Surplus optimization

To design the set of rules we want to implement it is important to understand the design goal (in this case surplus optimization). Understanding the problem helps in determining a useful set of rules. We assume that maximization of social surplus is the design goal. The general measure of social surplus for a specific pipeline \(j\) is :

\begin{equation} W(\boldsymbol {q},j) = \sum _i\bigl ( V_i(q_i,j) - P_i(\boldsymbol {q},j) \bigr ) \end{equation}

where :

\(q_i\)

is the share (quantity) of capacity that is allocated to firm \(i\),

\(\boldsymbol {q}\)

\((q_i, q_2 , \cdot \,)\), the vector of allocated capacity,

\(V_i(q_i, j)\)

is the value to firm \(i\) of capacity \(q_i\) on pipeline \(j\),

\(P_i(q_i, \boldsymbol {q}, j)\)

is the price firm \(i\) pays for capacity \(q_i\) on pipeline \(j\).

Specifically, in our experiments, the optimal surplus is the capacity allocation \(\boldsymbol {q^\ast }\) and the pipeline choice \(j^\ast \) that maximizes :

\begin{equation} \sum _{ij}\delta _j \bigl ( V_i(q_i, j) - P_i(q_i,\boldsymbol {q},j) \bigr ) \end{equation}

such that :

\begin{align} \sum _j \delta _j & \leq & 1, \\ \sum _{i} q_i & \leq & C_j \delta _j \quad \forall j, \\ \sum _{i} P_i(\boldsymbol {q}, j ) & \geq & T_j \delta _j \quad \forall j,\\ \delta _j&\in &\{ 0, 1 \} \quad \forall j. \end{align}

where :

\(\delta _j = 1\)

if pipeline \(j\) is allocated,

\(C_j\)

is the capacity of pipeline \(j\),

\(T_j\)

is the Total cost of constructing pipeline \(j\).

The initial research question posed is whether there exists a mechanism or set of institutional rules that can implement the maximal surplus allocation. In general, the answer is no NOTE: show proof.

We next consider the equilibrium outcomes of our proposed set of rules. NOTE show Nash Eq of proportional mechanism.