A Lost Opportunity Cost Model for Energy and Reserve Co-optimization Deqiang Gan, and Eugene Litvinov Abstract – The primary goal of this work is to investigate the basic energy and reserve dispatch optimization (cooptimization) in the setting of a pool-based market. Of particular interest is the modeling of lost opportunity cost introduced by reserve allocation. We derive the marginal costs of energy and reserves under a variety of market designs. We also analyze existence, algorithm, and multiplicity of optimal solutions. The results of this study are utilized to support the reserve market design and implementation in ISO New England control area.1 Index Terms – Electricity Market, Optimization, Power Systems, Spinning Reserve, Marginal Pricing I. NOMENCLATURE DP - node energy demand, a vector; DR e f - system reserve demand (or requirement), a scalar; - unit vector, every elements of e is unity; - lost opportunity cost function; F I l p/P r/R T γ ϕ - transmission thermal limit vector; - index set of generators (consists of 1, 2, 3, …., ); - lost opportunity cost price; - energy bid price/generation; - reserve bid price/allocation; - generation sensitivity factor matrix; - energy nodal price vector; - reserve clearing price. II. INTRODUCTION There exist two school of thoughts for market design as deregulation in power industry proceeds. They are commonly known as pool model and bilateral model [1,2]. ISO New England (ISO-NE) control area electricity market follows the concept of pool model [3,4]. In a pool-based market, energy and reserves are centrally and optimally allocated based on volunteer bids (in this paper reserve refers to 10-minute spinning reserve). Under twosettlement design [4], the allocation of energy and reserves are implemented in two steps, day-ahead scheduling and realtime dispatch. The optimization concepts of the two steps are, broadly speaking, similar. In this study, we focus on the market design issues in the real-time dispatch setting. While the idea of locational marginal pricing advocated in [5,6,7] seemingly dominates the pool-based energy markets, there is little consensus on how to structure reserve markets. In fact, the design of reserve markets is often a debatable topic. Discussions on this topic can be found in, say, [8-12, 22]. 1 The authors are with ISO New England, Inc., One Sullivan Road, Holyoke, MA, 01040, USA. Email: [email protected]; [email protected] One of the current major tasks in the ISO New England, Inc. is to study and possibly improve the existing reserve market. Aside from a long term forward market, which will not be discussed here, the following four alternative designs received attention in this effort: • Generators receive availability payments - Model (A); • Generators receive lost opportunity cost payments Model (L); • Generators receive availability and lost opportunity cost payments - Model (A+L); • Generators receive either availability or lost opportunity cost payment - Model (A|L). Availability payment refers to the payment to generators that offer reserve capability to the market. The definition of lost opportunity cost incurred in reserve allocation can be found in next section. More details about it can be found in New York, PJM, or New England ISO web site. The existing ISO-NE market, which is undertaking a major revision, follows Model (A+L). The revised market design will likely follow a variant of Model (A+L) as outlined in [23]. We will briefly discuss this revised market design subsequently. The three electricity markets in Northeastern USA all have or will have energy and reserve markets in which lost opportunity costs are explicitly compensated. From social welfare point of view, energy, reserve, and lost opportunity cost ought to be co-optimized. A systematic treatment on the formulation, solution algorithm, and pricing analysis of cooptimization becomes a timely issue. The primary goal of this work is to study the basic principles of co-optimization, laying down the engineering foundations of energy/reserve market design. The results can also be valuable for market implementation. We do not investigate how to choose the optimal market design which would require substantial economic analysis [13-15]. Of particular interest in this work is to present a general approach for modeling lost opportunity cost. In the context of co-optimization, we derive the formulae for calculating marginal costs under a variety of market designs. We also investigate such issues as solution existence, algorithm, and multiplicity of co-optimization. III. LOST OPPORTUNITY COST FUNCTIONS Suppose in a pool-based power market each generator submits a single bid block. The energy-only dispatch for one dispatch interval (5 minutes in the ISO-NE) can be expressed as [4-6]: Min p T P (0-1) P S .T . eT ( P − DP ) = 0 , T ( P − DP ) ≤ F , P ≤ P ≤ P . (0-2) The dispatch solution of the above model, let it be P̂ , does not need to respect reserve requirement. Some generators may have to be backed down to provide reserve. 0-7803-7519-X/02/$17.00 (C) 2002 IEEE The lost opportunity cost of a generator can be defined as follows: max{0, (γ − p )( Pˆ − P)}, γ > p f ( P, γ ) = . (1) γ≤p 0, The choice of specific form of f (⋅) is controversial. We consider two alternative cases: γ is a constant vector obtained from energy-only dispatch optimization (0-1)~(0-2); γ is the final energy price which itself is a variable of the optimization. For ease of exposition, the multi-block bids are not considered in this paper, the results could be easily extended to a multi-block case. IV. FORMULATION WHEN CONSTANT PRICE IS USED IN CALCULATING LOST OPPORTUNITY COST In this section, the lost opportunity cost function is assumed to take the following form: max{0, (γˆ − p )( Pˆ − P)}, γˆ > p, f ( P, γ ) = (2) γˆ ≤ p, 0, where γˆ is the constant energy price vector obtained from energy-only dispatch optimization (0). A. Optimization Formulation and Solution Existence Let us define a constant vector l as follows: γˆ − p, γˆ > p, l= γˆ ≤ p, 0, We have: f = max{0, l( Pˆ − P)} . (3) (4) The graph of a lost opportunity cost function is illustrated in Figure 2. Obviously, it is non-differentiable but it is continuous. f slope = - l P̂ P Figure 2. Lost Opportunity Cost Function when a Constant Price Is Used The problem of energy and reserve dispatch with consideration of lost opportunity cost is as follows: Min P,R S .T . T T p P + r R + ∑ fi T e ( P − DP ) = 0 , (5-1) (5-2) eT R = DR , (5-3) T ( P − DP ) ≤ F , (5-4) P≤ P+R≤ P , (5-5) f i = max{0, l i ( Pˆi − Pi )} , i ∈ I , P ≥ 0, R ≥ 0 . (5-6) (5-7) The above model corresponds to market design Model (A+L). To obtain the optimization model for market design Model (A), one only needs to assume that l = 0 in the above model. To obtain the optimization model for market design Model (L), one needs to assume that r = 0 in the above model. In the recent proposal [23], generators that are able to provide reserves are classified as Tier 1 and Tier 2 resources. In a nutshell, Tier 1 resources do not incur lost opportunity cost while Tier 2 resources do. The optimization formulation for Tier 1 resource is simply an energy-only optimization, and the optimization formulation for Tier 2 resources is similar to that of Model (A+L). By the continuity of objective function of the above problem, the answer to the question of solution existence is a qualified “yes”, provided the feasible set of the problem is nonempty [16, Weierstrass Theorem]. Now let us briefly discuss the optimization formulation for model (A|L). It can be stated as follows: Min p T P + ∑ [vi ri Ri + (1 − vi ) f i ] (6-1) S .T . e T ( P − DP ) = 0 , (6-2) P,R T e R = DR , (6-3) T ( P − DP ) ≤ F , (6-4) P≤ P+R≤ P , (6-5) f i = max{0, l i ( Pˆi − Pi )} , i ∈ I , P ≥ 0, R ≥ 0 , v ∈ {0,1} . (6-6) (6-7) The 0-1 variables v are introduced to enforce the condition that a generator either receives reserve availability payment or lost opportunity cost payment. Mathematically, this model is similar to Model (A+L), so we will not discuss it further. But the results to be presented can be extended to deal with this model. Before proceeding to study the solution algorithm, we note that locational reserve requirements are not discussed here but theoretically they could be incorporated into the presented framework without conceptual difficulty using methods presented in the literature [17,18]. B. Solution Algorithm This section suggests a solution algorithm for the optimization problem (5). First, let us convert the non- 0-7803-7519-X/02/$17.00 (C) 2002 IEEE differentiable lost opportunity cost function into a discrete function as follows: f i = ui l i ( Pˆi − Pi ) , ui ( Pˆi − Pi ) ≥ 0 , (1 − ui )( Pi − Pˆi ) ≥ 0 , u ∈ {0,1} With the above manipulations, the energy and reserve dispatch problem can be re-formulated as a 0-1 mixed integer programming problem as: ∂Γ ( ) = pi − l i − λ − T T µ − τ i + τ i = 0 , ∂Pi (9-1) ∂Γ ( ) = ri − ϕ − τ i + τ i = 0 . ∂Ri (9-2) The energy price is the marginal cost γ = λe + T T µ [5,6]. The reserve price is also set to the marginal cost ϕ . If the i-th generator does not have lost opportunity cost, we have u i = 0 , Pˆi − Pi < 0 . By the standard Kuhn-Tucker Min p T P + r T R + ∑ ui l i ( Pˆi − Pi ) (7-1) optimality theory, υ i = 0 . It follows that: S .T . e T ( P − DP ) = 0 , (7-2) eT R = DR , (7-3) T ( P − DP ) ≤ F , (7-4) P≤ P+R≤ P , u ( Pˆ − P ) ≥ 0 , (7-5) ∂Γ ( ) = pi − λ − T T µ − τ i + τ i = 0 , ∂Pi ∂Γ ( ) = ri − ϕ − τ i + τ i = 0 . ∂Ri P,R i∈I , ˆ (1 − ui )( Pi − Pi ) ≥ 0 , i ∈ I , P ≥ 0, R ≥ 0 , u ∈ {0,1} . i i (7-6) i (7-7) (7-8) The above 0-1 problem can be solved using the standard branch-and-bound method [19]. Note that the integer variables are associated with generators with reserve capabilities only. In ISO-NE, the number of generators is about 350, but the number of generators that have reserve capability is only about 50. Whether or not the standard branch-and-bound algorithm can meet the requirement of real-time application is out of the scope of this study. However, the formulation (7) allows us to investigate the pricing issues of energy and reserves. C. Optimality Conditions and Marginal Costs Suppose we find the optimal integer solution u * . Now it is straightforward to derive the marginal costs of energy and reserve. As usual, we form the Lagrangian function as follows: i∈I - λ [eT ( P − DP )] - ϕ (eT R − DR ) + µ T [T ( P − DP ) − F ] ( ) + τ T ( P − P − R) + τ T ( P + R − P ) - ∑ ω u * ( Pˆ − P ) - ∑υ (1 − u * )( P − Pˆ ) . (8) i i i i i∈I i i (10-2) The energy price is again the marginal cost γ = λe + T T µ . The reserve price is still given by ϕ . In the next section, we will present a pricing analysis based on equations (9)-(10). V. PRICING ANALYSIS FOR ALTERNATIVE MARKET DESIGNS In this section, we present a pricing analysis for each of the models mentioned in the previous section. Whenever possible, we will indicate if there exist multiple solutions. A. Model (A) Under this simple model, l = 0 . The reserve clearing price equals to the reserve availability price of the most expensive generator that is designated to supply reserve. When the bid prices for reserves of many generators are all equal to zero (this happens quite often), then ϕ = 0 and generally there are multiple solutions in reserve allocation. In fact, there is a continuum of reserve solutions. For instance, consider the following problem: Min 10 PA + 20 PB + 0 R A + 0 RB PA + PB = 70 , R A + RB = 20 , 0 ≤ PA + R A ≤ 100 , PA ≥ 0 , R A ≥ 0 , 0 ≤ PB + RB ≤ 100 , PB ≥ 0 , RB ≥ 0 , The optimal energy dispatch is PA = 70, PB = 0 , but any reserve dispatch that meet R A + RB = 20 is an optimal reserve dispatch. To find an unique solution, one method is to designate reserve contributions to the generators with lowest energy bid prices. S .T . Γ = p T P + r T R + ∑ ui*l i ( Pˆi − Pi ) i∈I (10-1) i i If the i-th generator has lost opportunity cost, then we have that ui = 1 , Pˆi − Pi > 0 . By the standard Kuhn-Tucker optimality conditions, we have that ω i = 0 , and: B. Model (L) Under this model, r = 0 . Let us consider two situations. In the first situation, the optimal dispatch does not require 0-7803-7519-X/02/$17.00 (C) 2002 IEEE paying generators lost opportunity cost. This is just the situation in model (A) when the bid prices for reserves are all equal to zero. So it is quite possible that there are multiple solutions for reserve allocation. To resolve this problem, again, one could designate reserve contributions to generators with lowest energy bid prices. Now let us consider the second situation where, in the optimal dispatch, some generators are paid lost opportunity cost. In the sequel we show that the marginal cost of producing reserves, ϕ , can still be greater than zero. To get a feel of what ϕ could be, let us derive an alternative expression of ϕ from equations (9-1)-(9-2): ϕ = ri + l i + γ i − pi (11) When there is no congestion, it is obvious that γ i ≥ γˆi . Since r = 0 , ϕ = l i + γ i − p i ≥ 2l i > 0 (12) Recall that the marginal cost of a product is equal to the change of production cost as the demand increases by a small amount [20]. Based on this principle, let us verify the result (12). Price Suppose reserve requirement is increased by a infinitesimal ε . The change of total production cost would consist of three components: • • • Energy cost increase of generator B, which is p B ε ; Energy cost decrease of generator A, which is − p Aε ; Lost opportunity cost increase of generator A, which is ( p B − p A )ε . Note that p B happens to be equal to the clearing price. The change of production cost would be p B ε − p Aε + ( p B − p A )ε = 2l Aε . This indicates that the reserve marginal cost is equal to 2l A which is consistent with the result in (12). C. Model (A+L) This model, which is being used in the existing ISO-NE market, is quite similar to Model (L) except that the problem of solution multiplicity in allocating reserves does not happen very often. The reason is that ISO has availability bids. Following the observation derived from equations (11)-(12), if generator i is designated to supply reserve and it has lost opportunity cost, we have: Load ϕ = ri + l i + γ i − pi ≥ ri + 2l i . D. Model (A|L) clearing price A B Quantity Figure 3. Clearing Energy-only Market in a 5-Generator System Price Load The pricing analysis for this model is similar to that of model (A+L) so we will not proceed further. Under such a model, there can still exist multiple solutions in reserve allocation, but this does not happen very often because ISO has availability bids. F. A Variant of Model (L) This model is the same as Model (L) except that lost opportunity cost is not explicitly included in the objective function (but generators do receive lost opportunity cost payment). The optimization formulation is as follows: Min pT P (14-1) S .T . e T ( P − DP ) = 0 , (14-2) P clearing price T X (13) A B Quantity Figure 4. Clearing Energy and Reserve Market in a 5Generator System (the shaded area is designated for reserve) e R = DR , (14-3) T ( P − DP ) ≤ F , (14-4) P≤ P+R≤ P , P ≥ 0, R ≥ 0 . (14-5) (14-6) When there is no congestion, the solution of the above problem coincides with the solution of the problem where lost opportunity cost is explicitly included in objective 0-7803-7519-X/02/$17.00 (C) 2002 IEEE function! This can be illustrated using Figure 4 in which generator A or X can provide reserve but they also demand lost opportunity cost. Based on formulation (14), generator A would be designated to provide reserve and it would be paid lost opportunity cost. Apparently, if lost opportunity cost is included explicitly into (14-1), the optimal solution would still be the same! As in the model (L), there are multiple solutions for reserve allocation when there is ample reserve capacity margin. VI. FORMULATION WHEN VARIABLE PRICE IS USED IN CALCULATING LOST OPPORTUNITY COST In this section we assume that lost opportunity cost function takes the most general form (1). This optimization is self-referential because the lost opportunity cost depends on variable price γ which is not known until the final solution is obtained! In this section we suggest a method to get around this impasse. Let us suppose temporarily that the reserve dispatch is given, let it be R * . Then it is trivial to compute optimal energy dispatch and resultant energy prices: Min pT ⋅ P (15-1) S .T . e T ( P − DP ) = 0 , (15-2) P T ( P − DP ) ≤ F , * (15-3) * P−R ≤ P≤ P −R . (15-4) The Lagrangian function of the above optimization problem is easily obtained as: Γ = p T P + λ [ eT ( DP − P) ] + µ T [ F − T ( P − DP )] ( ) (16) + τ T ( P − R* − P) + τ T ( P + R* − P ) . The spot prices are given by γ = λe + T T µ . The key to solving the problem is to find out an energy dispatch that is optimized taking into account reserve requirements. Consider the following bi-level optimization problem: Min pT P + r T R + ∑ f i (17-1) S .T . eT R = DR , R ≥ 0 , (17-2) R max{0, (γ i − pi )( Pˆi − Pi )}, γ i > pi , i∈I , fi = γ i ≤ pi 0, (17-3) T (17-4) Min p ⋅ P P S .T . e T ( P − DP ) = 0 , (17-5) T ( P − DP ) ≤ F , (17-6) P−R≤ P≤ P −R, (17-7) P≥0. (17-8) In the above formulation, R is a variable in the upperlevel optimization and a parameter in the lower-level optimization. P stays in the lower-level optimization subproblem. The primal solution of the above bi-level optimization problem is the desired optimal dispatch of energy and reserves. The dual solution contains locational marginal prices. In what follows we describe briefly how to solve the bilevel optimization problem (17). By the standard theory of linear programming, the primal and dual solutions of the lower-level sub-problem depends on the optimal basis, B , of the lower-level sub-problem. That is: P = B −1b( R ) , (18-1) λ T −1 = (B ) p , µ (18-2) where b(R) is the right-hand-side of the lower-level subproblem, it is a linear function of R . To attack the bi-level optimization problem, observe that λ and µ do not depend on the specific value of R, rather they depend on which constraints are included in optimal basis only. As a result, suppose temporarily that the optimal basis B is known, then energy prices are known. Let it be γ~ , ~ and l be as defined based on equation (3). Now the bi-level optimization problem can be cast as: Min pT P + r T R + ∑ f i (19-1) S .T . eT R = DR , R ≥ 0 , ~ f i = max{0, li ( Pˆi − Pi )} , i ∈ I , (19-2) R −1 P = B b( R ) . (19-3) (19-4) The above problem possesses a structure that is similar to that of (5). It can be solved using the general algorithm described in Section IV-B. The question remains to be how to find out the optimal basis. One obvious solution is to enumerate all the combinations of bases of lower level sub-problem. For example, in the 5-generator system illustrated in Figure 3 where each generator is required to submit single block bid, there are only five possible energy prices. A better solution is to apply a standard branch-and-bound algorithm [19]. Since branch-and-bound algorithm is fairly familiar to the power engineering audience, we will not discussed it here. The readers are referred to [19] for details. VIII. FINAL REMARKS In this paper we studied four alternative energy/reserve market designs that received attention in ISO-NE. We presented a fairly detailed analysis on the basic formulation, 0-7803-7519-X/02/$17.00 (C) 2002 IEEE solution algorithm, and pricing formulae of co-optimization under these market designs. The results of the research have been utilized to support, from engineering perspective, the reserve market design and implementation in ISO-NE. The main finding is that energy, reserve, and lost opportunity cost co-optimization is, in general, a non-differentiable and possibly bi-level optimization problem. This problem can be further converted into a mixed integer programming problem. A standard algorithm for solving these problems is that of branch-and-bound. This algorithm can be efficient or exceedingly slow, depending upon the size of the problem. Whether or not a standard branch-and-bound algorithm can meet engineering requirement is thus a subject of additional research. IX. ACKNOWLEDGEMENT The authors have benefited from the discussions in ISONE Reserve Market Design Working Group. The opinions described in the paper do not necessarily reflect those of ISO New England, Inc. The authors remain solely responsible for errors. X. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] H. Singh, S. Hao, A. Papalexopoulos, “Transmission Congestion Management in Competitive Electricity Markets”, IEEE Trans. On Power Systems, vol. 13, no. 2, May 1998, pp. 672-680 F. F. Wu, P. Varaiya, “Coordinated Multilateral Trades for Electric Power Networks: Theory and Implementation”, International Journal of Electrical Power & Energy Systems, vol. 21, no. 2, 1999 K. W. Cheung, P. Shamsollahi, D. Sun, J. Milligan, M. Potishnak, "Energy and Ancillary Service Dispatch for the Interim ISO New England Electricity Market", IEEE Trans. on Power Systems, vol. 15, no. 3, August 2000, pp. 968-974. D. Gan, Q. Chen, “Locational Marginal Pricing: New England Prospective”, presentation summary for Panel Session “Flow-gate and Location Based Pricing Approaches and Their Impacts”, in Proceedings of IEEE PES Winter Meeting, Columbus, Ohio, USA, January 2001. M. C. Caramanis, R. E. Bohn, F. C. Schweppe, “Optimal Spot Pricing: Practice and Theory”, IEEE Trans. On Power Apparatus and Systems, vol. 101, no. 9, 1982 M. L. Baughman, S. N. Siddiqi, “Real Time Pricing of Reactive Power: Theory and Case Study Results”, IEEE Transactions on Power Systems, vol. 6, no. 1, 1991, pp. 23-29 W. W. Hogan, “Contract Networks for Electric Power Transmission”, Journal of Regulator Economics, vol. 4, 1992, pp. 211-242 D. Chattopadhyay, B. B. Chakrabarti, E. G. Read, “Pricing For Voltage Stability”, in Proceedings of PICA, Melbourne, Australia, May 2001 E. H. Allen, M. D. Ilic, “Reserve Markets for Power Systems Reliability”, IEEE Trans. on Power Systems, vol. 15, no. 1 , 2000, pp. 228–233 H. Singh, A. Papalexopoulos, “Competitive Procurement of Ancillary Services by an Independent System Operator”, IEEE Trans. On Power System, vol. 14, no. 2, 1999, pp. 498-504 M. Flynn, W. P. Sheridan, J. D. Dillon, M. J. O'Malley, “Reliability and Reserve in Competitive Electricity Market Scheduling”, IEEE Trans. on Power Systems, vol. 16, no. 1 , 2001, pp 78 –87. X. Ma, D. Sun, K. Cheung, “Energy and Reserve Dispatch in a Multizone Electricity Market”, IEEE Transactions on Power Systems, vol. 14, no. 3 , 1999, pp. 913-919 D. Gan, D. V. Bourcier, “Locational Market Power Screen and Congestion Management: Experience and Suggestions”, IEEE Trans. on Power Systems, accepted for publication R. J. Green, “Competition in Generation: The Economic Foundations”, Proceedings of the IEEE, vol. 88, no. 2, February 2000, pp. 128-139 [15] R. Zimmerman, R. Thomas, D. Gan, C. Murillo-Sanchez, “A WebBased Platform for Experimental Investigation of Electric Power Auctions”, Decision Support Systems, vol. 24, no. 3&4, January 1999, pp. 193-205 [16] M. S. Bazaraa, H. D. Sherali, C. M. Shetty, Nonlinear Programming – Theory and Algorithms, John Wiley & Sons, Second Edition, 1993 [17] S. Hao, D. Shirmohammadi, “Clearing Prices Computation for Integrated Generation, Reserve and Transmission Markets”, in Proceedings of PICA, Sydney, Australia, May 2001 [18] M. Aganagic, K. H. Abdul-Rahman, J. G. Waight, “Spot Pricing of Capacities for Generation and Transmission of Reserve in An Extended Poolco Model”, IEEE Trans. On Power Systems, vol. 13, no. 3, August 1998, pp. 1128-1134 [19] J. F. Bard, Practical Bilevel Optimization – Algorithms and Applications, Kluwer Academic Publishers, Dordrecht, The Netherlands, 1998 [20] P. R. Gribik, G. A. Angelidis, R. R. Kovas, “Transmission Access and Pricing with Multiple Separate Energy Forward Markets”, IEEE Trans. On Power Systems, vol. 14, no. 3, 1999, pp. 865-876 [21] A. Jayantilal, K. W. Cheung, P. Shamsollahi, F. S. Bresler, “Market Based Regulation for the PJM Electricity Market”, In Proceedings of PICA, Sydney, Australia, May 2001 [22] N. S. Rau, “Optimal Dispatch of a System Based on Offers and Bids A Mixed Integer LP Formulation”, IEEE Transactions on Power Systems, vol. 14, no. 1, 1999, pp. 274 –279 [23] PJM Interconnection, L.L.C, “Spinning Reserve Market Business Rules”, July 16, 2001 [Online]. Available: http://www.pjm.com Deqiang Gan is currently a Senior Analyst in ISO New England, Inc. where he works on issues related to the design, implementation, and economic analysis of electricity markets. Prior to joining ISO New England, Inc., he held research positions at several universities. Deqiang received a Ph.D. in Electrical Engineering from Xian Jiaotong University, China, in 1994. Eugene Litvinov obtained the BS and MS degree from the Technical University in Kiev, Ukraine, and Ph.D. from Urals Polytechnic Institute in Sverdlovsk, Russia. He is currently a Director of Technology in the ISO New England. His main interests are power system market clearing models, system security, computer applications to power systems, information technology. 0-7803-7519-X/02/$17.00 (C) 2002 IEEE
© Copyright 2026 Paperzz