1. What kind of problems do you usually face in your energy modeling work? ,2. What type of optimization problems do you usually solve? ,3. What energy system modeling framework do you primarily use? ,4. Which solvers do you typically use for your energy modeling tasks?,5. What are the most common algorithms/options you use when solving?,6. What are the most common problems you encounter when using solvers for energy modeling?,"7. On average, how long does it take to solve a typical problem of your chosen size using your chosen solver? ",8. Scale of typical problems (max. size of initial LP problem given to solver):,"9. What do you usually do when the solver cannot solve a model? Please rank:
(1 = first thing you try, 4 = last thing you try) [Try another solver or algorithm]","9. What do you usually do when the solver cannot solve a model? Please rank:
(1 = first thing you try, 4 = last thing you try) [Use more powerful hardware]","9. What do you usually do when the solver cannot solve a model? Please rank:
(1 = first thing you try, 4 = last thing you try) [Reduce the model size]","9. What do you usually do when the solver cannot solve a model? Please rank:
(1 = first thing you try, 4 = last thing you try) [Other (please specify below)]",(Other) What do you do when the solver cannot solve a model?,10. Why did you choose your current solvers?, 11. Do you primarily choose solvers based on their performance in handling large-scale problems?,12. Would you prefer to use a different solver if it offered better performance or features? ,13. Would you be willing to pay licence fees to use a proprietary solver instead of an open-source solver if it offers better performance/features?,14. How do you evaluate the performance of a solver before deciding to use it? ,15. What improvements or features would you like to see in future solver developments to better meet your energy modeling needs? ,16. Can you share an example of a particularly challenging problem you solved and how you chose the solver for it? ,17. Can you describe any specific criteria or benchmarks you use to evaluate and compare different solvers? ,18. Can you recommend any other benchmarking websites? (not necessarily for solvers/energy models) Please also explain why you like it.,19. In what field do you work?
"Economic dispatch, Linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",oemof,"Gurobi, CBC",Solver default,High computational time,3-6 hours,,4,3,4,4,"Debug the model - check constraints, check for contradicting assumptions..

(BTW, the rank of the question above is not clear to me, I assume 4 is high)",Recommended by peers,Yes,Yes,Maybe,Benchmark tests,I need a very fast open source solver for ad-hoc/live optimization in web apps,,,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning",Linear programming (LP),PyPSA,Gurobi,Solver default,"High computational time, Solver convergence issues",3-6 hours,30-100 million,,,4,,,Best performance for specific models,Yes,Yes,No,User reviews and feedback,It should take less computational time and should be free.,I,,,"Academic Institution (University, College, Research Institute)"
Unit commitment,"Mixed Integer Linear Programming (MILP), Meta-heuristics (usually implemented ad hoc), mat-heuristics",.,"Gurobi, CPLEX","Solver default, Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm","High computational time, Memory limitations, License costs",3-6 hours,More than 100 million,4,1,2,1,"Decomposition, heuristics ",Industry standard,Yes,Yes,Maybe,Benchmark tests,Comparable performances w/ state of the art solver ,"https://www.roadef.org/challenge/2010/en/

I admit it was the last time I worked on an energy related problem, so you may want to treat my reply as an outlier;)","I cannot recall them, but there are some publicly available benchmarks which are periodically updated",(see previous answer),"Private Sector (Industry, Business)"
Power system planning,Linear programming (LP),MicroGridsPy,"Gurobi, Highs","Solver default, Primal or dual simplex",Model infeasibility,Less than 1 hour,Below 1 million rows / columns,2,1,4,4,Edit the model inputs and equations,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,Open source solver ,,,,"Academic Institution (University, College, Research Institute)"
"Unit commitment, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",highRES and TIMES,CPLEX,"Interior Point Method (IPM), or Barrier algorithm",High computational time,1-3 hours,,,4,,,,"legacy, standard for use with GAMS for TIMES",Yes,Yes,Yes,Benchmark tests,Open source solvers capable of handling larger problems and delivering solve times similar to CPLEX,I essentially always use CPLEX,I've looked at the Openmod forum threads comparing HiGhs with others ,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",Developed in-house model,"Highs, CBC, SCIP",Solver default,"High computational time, Model build time",Less than 1 hour,Below 1 million rows / columns,3,4,1,2,"Decompose the problem, typically in time dimension, ie moving horizon approach or similar.",Open-source licensing,No,Maybe,Maybe,In-house performance benchmarks on own data,Infeasibility analysis (output conflicts).,,"Open-source licensing. Availability of modeling interfaces (eg Python library), allowing user callbacks.","Mittelmann benchmarks for LP and MIP problems are most widely used, but solvers maybe too closely tuned to these benchmark instances.","Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",Developed in-house model,Gurobi,Running multiple solver instances in parallel,"High computational time, Model infeasibility",1-3 hours,1-30 million,3,2,1,,,Industry standard,No,Yes,Yes,Benchmark tests,,,,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Power system planning, Portfolio optimization (price taking models)","Linear programming (LP), Mixed Integer Linear Programming (MILP)","Commercial software (PLEXOS, SAInt)","Gurobi, CPLEX",Solver default,Model infeasibility,3-6 hours,,2,1,3,,,Industry standard,Yes,Yes,No,Benchmark tests,,,,,"Private Sector (Industry, Business)"
Power system planning,Non-linear,PyPSA,Highs,"Solver default, Interior Point Method (IPM), or Barrier algorithm",High computational time,3-6 hours,Below 1 million rows / columns,3,2,1,,,already known,Yes,Yes,No,Being trained at school / uni with this solver,,,,,"Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",Calliope,"Gurobi, CPLEX","Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","Memory limitations, Solver convergence issues",3-6 hours,1-30 million,3,4,2,,Simplify the model,Best performance for specific models,Yes,Yes,Yes,User reviews and feedback,,,,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Non linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear",GridCal,"Gurobi, CPLEX, Highs, CBC","Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","High computational time, Memory limitations, Model infeasibility",1-3 hours,30-100 million,,,,4,GridCal automatically relaxes and minimizes the problematic constraints and (almost) always solves after that,Best performance for specific models,Yes,Yes,Yes,Benchmark tests,Faster handling of integer vars in OS solvers and automatic decomposition,,Unit commitment with grid in a real model (i.e several countries real grid),,"Private Sector (Industry, Business)"
Multi energy system planning,Linear programming (LP),PyPSA,Gurobi,"Solver default, Interior Point Method (IPM), or Barrier algorithm","High computational time, Memory limitations, Solver convergence issues, Model infeasibility",1-3 hours,30-100 million,3,1,2,,,Recommended by peers,Yes,Yes,Maybe,User reviews and feedback,GPU parallelization,Gurobi helped with discovering infeasibilities,Just testing runtime on my model,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Non linear optimal power flow, Power system planning","Mixed Integer Linear Programming (MILP), Non-linear","PowerModels, PowerModelsDistribution","Highs, CBC, GLOP, Juniper, Ipopt","Solver default, Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm, branch-and-bound/cut/price variants","High computational time, Solver convergence issues",Less than 1 hour,1-30 million,1,4,2,,,Best performance for specific models,Yes,Yes,Maybe,my own benchmark problems,more advanced presolvers for nonlinear problems,,"I'm interested in learning impedance values of networks from measurements, using multiperiod OPF models with variable impedances. ",,"Private Sector (Industry, Business)"
Linear optimal power flow,"Linear programming (LP), Mixed Integer Linear Programming (MILP)","Casadi, cvxpy","Gurobi, Highs, Clarabel","Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm","High computational time, Memory limitations",Less than 1 hour,Below 1 million rows / columns,1,3,1,,,Recommended by peers,No,Yes,Maybe,Benchmark tests,"There are very good LP/QP solvers, but MILP solvers are currently lacking in performance.",,,,"Private Sector (Industry, Business)"
Economic dispatch,Linear programming (LP),PyPSA,Highs,Solver default,High computational time,Less than 1 hour,1-30 million,4,1,3,,,Recommended by peers,Yes,Yes,Yes,Benchmark tests,Open source barrier solver please,Many timestep PyPSA dispatch optimization,"The latest PyPSA model I have, time it",,"Private Sector (Industry, Business)"
Multi energy system planning,"Linear programming (LP), Mixed Integer Linear Programming (MILP)",PyPSA,"Gurobi, Highs, GLPK, CBC, SCIP",Solver default,"High computational time, Model infeasibility",1-3 hours,1-30 million,1,3,2,4,Reformulate model if possible,Industry standard,Yes,Yes,Maybe,Benchmark tests,Out of the box installation in Windows + Python ,MILP for District Heating with Storage,,"https://plato.asu.edu/bench.html transparency, updates, simplicity","Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment",Mixed Integer Linear Programming (MILP),Developed in-house model,"CPLEX, Highs, CBC",Solver default,"High computational time, Model infeasibility",Less than 1 hour,Below 1 million rows / columns,1,1,4,2,Simplify the model,Industry standard,Yes,Maybe,Yes,Benchmark tests,"Transparent, by default, easy to use decomposition mechanism",,,,"Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",Developed in-house model,Gurobi,"Interior Point Method (IPM), or Barrier algorithm","High computational time, Solver convergence issues",Less than 1 hour,30-100 million,2,2,1,,,"Tested different ones, this one works the best.",Yes,Yes,Maybe,Tests with my own model.,Would love (of course) for one of the open-source ones to perform as good as commercial ones.,,I just try to see which one works best with my models.,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Linear optimal power flow, Power system planning, Multi energy system planning",Linear programming (LP),PyPSA,Highs,"Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm","High computational time, Solver convergence issues",More than 6 hours,1-30 million,2,1,4,1,,free,Yes,Yes,Maybe,Benchmark tests,Faster LP solving time. More heuristics to help convergence,pypsa with lots of nodes. Chose highs as it was the best free option,,,"Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear",Developed in-house model,knitro,Solver default,High computational time,1-3 hours,1-30 million,3,1,1,2,,Recommended by peers,Yes,No,Yes,Benchmark tests,,,,,Non-Profit Organization
"Economic dispatch, Unit commitment, Non linear optimal power flow, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear, multi-objective",proprietary,InsideOpt Seeker,"Solver default, Seeker self tunes its algorithm for the problem","High computational time, Model infeasibility",Less than 1 hour,Below 1 million rows / columns,4,2,3,1,Remodel the problem,Best performance for specific models,No,Maybe,Yes,test on problem to be solved,"Ability to model the actual problem as accurately as possible. this means non-linear, non-convex, non-differentiable multi-objective optimization.",Unit commitment for a network of hydropower plants,solution quality after one hour,,"Private Sector (Industry, Business)"
"Economic dispatch, Multi energy system planning, Market simulations","Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear, + MIQLP",Developed in-house model,"Gurobi, PATH",Solver default,"High computational time, Model infeasibility",Less than 1 hour,,3,1,4,4,"Try to reformulate the problem, use heuristics/machine learning or simulations.",Best performance for specific models,Yes,Yes,Maybe,Comparing its performance and results with the current solver I am using.,"More clarity on the algorithms used, and information about why (many modellers do not have a really clear idea of the algorithms used by their solvers). Option to choose a certain algorithm (maybe already doable, didn't really dig into that).","I am working with risk-averse actors of de-centralised problems. I used to like to formulate an upper level government problem with the lower level problem optimality conditions as constraints. Those types of problems are non-convex. I chose to go for some decomposition (ADMM) methods, and agent-based modelling instead of optimization.","I mostly focus on the tractability, computational time and results in my case.",,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Power system planning, Multi energy system planning",Mixed Integer Linear Programming (MILP),MiEAP - Mipower Energy Action Planning,"GLPK, CBC","Solver default, Running multiple solver instances in parallel",High computational time,1-3 hours,More than 100 million,2,3,1,,,Industry standard,Yes,Yes,No,Benchmark tests,,,,,"Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Linear optimal power flow","Linear programming (LP), Mixed Integer Linear Programming (MILP)",ASSUME,"GLPK, CBC",Solver default,Solver convergence issues,Less than 1 hour,Below 1 million rows / columns,4,2,4,4,debug my MILP,"open-source, no license hassle",No,Yes,No,"taking the low hanging fruit, and sticking until needed otherwise",,,,I like https://www.gsmarena.com/ and https://cpu.userbenchmark.com/ - they provide good filters and give a clear idea of pros and cons and link to other benchmarks,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",PyPSA,"CPLEX, Highs","Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel",High computational time,1-3 hours,1-30 million,3,1,2,2,"There are tricks with chopping up the model either into multiple time horizons, or by separating investment decisions in for instance generation capacity and transmission capacity into separate solver stages - you can lose important details, but still gain key insight into system behaviour. ","Highs is chosen for being open source MIT licensed while still being very perfomant even on quite large relevant problems. When highs doesn't cut it, we sometimes bring cplex into the mix, though the licensing makes certain valuable use cases impossible with the commercial solver. ",Yes,Maybe,Maybe,"Basically run my own tests on my own ""real world"" problems ","Ideally, highs would reach a stage where it performs similarly to cplex in single threaded mode. Currently, highs tends to be about ten times slower than cplex, though running cplex single threaded shrinks the gap to be on the order of cplex being twice to four times faster. It all depends on problem and solver algo, of course, but rough numbers from experience. ","Classic problem that in practice is impossible to solve at the moment, without resolving to cplex, is unit commitment of diesel generators to represent power dependent efficiency curve in the same model where investment in battery energy storage is relevant. Even cplex struggles significantly, which makes sense, since UC plus investment is a pain in the ass. Have only recently heard about linearized UC approach that sounds interesting to test. ",,,"Private Sector (Industry, Business)"
"Linear optimal power flow, Power system planning",Linear programming (LP),TEMOA,"Gurobi, CPLEX","Solver default, Primal or dual simplex","High computational time, Model infeasibility",Less than 1 hour,Below 1 million rows / columns,2,3,4,,,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,,,,,Non-Profit Organization
"Economic dispatch, Unit commitment, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",PyPSA,"CBC, Xpress","Interior Point Method (IPM), or Barrier algorithm, IPM for capacity expansion without crossover",High computational time,1-3 hours,1-30 million,3,1,1,,Reduce model size,Industry standard,Yes,Yes,Yes,Testing it on trial basis,Commercial solvers are just so much faster,"Capacity expansion planning for South African power system 2025 to 2050 (perfect foresight, but single node)",Time to solve on own problems,Hans Mittelman,"Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",PLEXOS,"Gurobi, CPLEX, Highs",Solver default,"High computational time, Memory limitations, Model infeasibility",1-3 hours,1-30 million,4,3,2,1,"Usually caused by infeasibilities, so looking for the root cause of the infeasibility",Best performance for specific models,Yes,Yes,Yes,Benchmark tests,"Inclusion of stochastic problems to address intermittent renewable generation, uncertainties on loads and climate change",,,,"Private Sector (Industry, Business)"
"Unit commitment, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)","Dolphyn, OMNI-ES, oemof","Gurobi, CPLEX, Highs","Interior Point Method (IPM), or Barrier algorithm","High computational time, Memory limitations",1-3 hours,30-100 million,2,1,3,4,Scale the model better,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,,,,,"Academic Institution (University, College, Research Institute)"
Civilization viability,Dynamic systems analysis,Mass and energy flows How the world works v2.0,Graphical flow models,Extended space and time boundaries,Missing causality (and sinks and sources data,More than 6 hours,Below 1 million rows / columns,,,2,1,I agrigate parameters,The other tools were timeblind,Yes,Maybe,No,Is it capable of including causal relationships. ,Solutions over time to end of this century ,https://www.youtube.com/watch?v=Cnyweoke5Cc,Who gets injured and when,Www.skil.org,Knowledge integration lab. Www.skil.org
"Economic dispatch, Unit commitment, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",PyPSA,"CPLEX, CBC",Solver default,"High computational time, Model infeasibility",3-6 hours,30-100 million,3,2,1,4,,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,Display which variables are causin unfeasibilities,,,,Government Agency
Power system planning,Linear programming (LP),PREP-SHOT,"Gurobi, Highs, Mosek and COPT","Solver default, Running multiple solver instances in parallel, Benders decomposition","High computational time, Memory limitations",More than 6 hours,More than 100 million,4,,4,,,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,No idea,Using Bender decomposition skills to decompose large-scale model.,solving time and gap,No idea,"Academic Institution (University, College, Research Institute)"
"Linear optimal power flow, Non linear optimal power flow, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",pyomo,"Highs, GLPK, CBC",Primal or dual simplex,"High computational time, Memory limitations, Model infeasibility",Less than 1 hour,Below 1 million rows / columns,2,4,3,,,Recommended by peers,Yes,Yes,Maybe,User reviews and feedback,,,,,"Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Non linear optimal power flow, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",powersystems.jl / proper code,"Gurobi, Highs","Solver default, Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","High computational time, Memory limitations, Solver convergence issues, Model infeasibility",1-3 hours,1-30 million,3,2,4,1,"implement decomposition methods (Benders, Lagrangian relaxation)",Best performance for specific models,Yes,Maybe,No,Benchmark tests,increase performance in large scale problems,large scale deterministic unit commitment problems - gurobi was chosen due to the performance and free license for university.. currently i'm changing some problems to HiGHS,computational time / optimality gap for different unit commitment problems ,n/a,"Academic Institution (University, College, Research Institute)"
"Unit commitment, Power system planning",Mixed Integer Linear Programming (MILP),Custom,"Gurobi, SCIP (for tests), Glop (for tests), Mosek (benchmarking)",Primal or dual simplex,High computational time,1-3 hours,More than 100 million,,,,1,Decomposition techniques,Best performance for specific models,Yes,Yes,Yes,Benchmark tests,,,,,"Private Sector (Industry, Business)"
"Linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",GenX,"Gurobi, CPLEX, Highs","Interior Point Method (IPM), or Barrier algorithm, Parallelized custom Benders decomposition method","High computational time, Memory limitations, Numerical stability issues",More than 6 hours,30-100 million,2,,1,3,Try decomposition methods,Best performance for specific models,Yes,Yes,Yes,Being trained at school / uni with this solver,"Improved open source solvers that are specialized in sparse large-scale (millions of variables/constraints) linear programs (e.g. very good interior point, resolution of numerical instability issues).",We work on large-scale planning models with 10s of millions of variables/constraints and have applied custom decomposition methods and/or time sampling methods with Gurobi to solve large-scale problems.,Performance with large-scale capacity expansion planning problems.,No.,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",GenX,Gurobi,"Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm",High computational time,More than 6 hours,30-100 million,3,1,2,,,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,Good warmstarting,,,,"Academic Institution (University, College, Research Institute)"
"Unit commitment, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",PyPSA,Highs,"Solver default, Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm","High computational time, Memory limitations",3-6 hours,1-30 million,1,2,3,,,Industry standard,Yes,Yes,Maybe,Benchmark tests,,,,,Non-Profit Organization
"Linear optimal power flow, Non linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",Developed in-house model,"Highs, SCIP",Solver default,High computational time,3-6 hours,Below 1 million rows / columns,4,1,4,4,Design custom algorithms build on solver smaller sizes models,Best performing open source solvers,Yes,Yes,No,Benchmark tests,,,,Mittelmann benchmarks https://plato.asu.edu/bench.html,"Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",PyPSA,"Gurobi, CPLEX, Highs, GLPK, CBC, COPT, MOSEK, FICO XPRESS, MINDOPT, SCIP","Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm","High computational time, Memory limitations, Solver convergence issues",More than 6 hours,1-30 million,4,1,2,3,Tweaking numeric solver parameters; problem scaling,Best performance for specific models,Yes,Yes,No,Benchmark tests,"GPU usage, parallelisation",,"speed, memory, stability",https://plato.asu.edu/bench.html,"Academic Institution (University, College, Research Institute)"
"Unit commitment, Linear optimal power flow, Non linear optimal power flow, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",PyPSA,"Gurobi, Highs","Solver default, Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","High computational time, Solver convergence issues, Model infeasibility",1-3 hours,1-30 million,2,1,1,,,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,,,,,Non-Profit Organization
"Economic dispatch, Linear optimal power flow, Power system planning, Multi energy system planning",Linear programming (LP),PyPSA,Gurobi,"Solver default, Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm","High computational time, Memory limitations",1-3 hours,1-30 million,4,1,2,3,Try to reformulate the constraints ,Industry standard,Yes,Yes,Maybe,User reviews and feedback,,,None,Not really,"Academic Institution (University, College, Research Institute)"
"Linear optimal power flow, Non linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",Pandapower,"Gurobi, CPLEX, Highs","Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","High computational time, Memory limitations, Solver convergence issues, Model infeasibility",More than 6 hours,1-30 million,1,3,2,4,"other solver options, decompositions, linearizations, different modelling, check numerics, etc.",Best performance for specific models,Yes,Yes,Yes,Free testing license,,,,https://plato.asu.edu/bench.html,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Power system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",MATCH/SWITCH,"CPLEX, CBC","Solver default, Running multiple solver instances in parallel",High computational time,More than 6 hours,1-30 million,2,3,1,,,Cost,No,Yes,Maybe,Benchmark tests,,,,,"Academic Institution (University, College, Research Institute)"
market interventions by governments,mixed complementarity modeling,GAMS,Path,complementarity algorithms,Model infeasibility,Less than 1 hour,Below 1 million rows / columns,,,,1,Find what is mismatched in the primal/dual formulation,It is the only one for these problems,No,Yes,Yes,experience,,A model of crude flows and tanker markets with controls on Russian prices,Path is the only solver I know for these problems,no,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear",Dolphyn,"Gurobi, Highs","Interior Point Method (IPM), or Barrier algorithm",Memory limitations,3-6 hours,More than 100 million,3,1,2,,,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,"I would like to more easily build / solve models in Float32 precision, so that I can represent larger systems",We have multi-decade hourly resolution models of power and multi-sector models. We have tested solving sub-sets of the problem using several solvers. Gurobi has the best run time by approximately an order of magnitude and the memory required is similar.,,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Linear optimal power flow, Multi energy system planning",Linear programming (LP),oemof,CBC,Solver default,"High computational time, Memory limitations",Less than 1 hour,Below 1 million rows / columns,4,3,2,1,Cange problem to find problematic issue,"free to use, easy to apply with oemof",No,Yes,No,User reviews and feedback,Modelling to generate alternatives / pareto fronts,/,/,/,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning",Linear programming (LP),"Switch, energyRt, PyPSA (3 Modeling Framework)","CPLEX, Highs, GLPK, CBC","Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm","High computational time, Solver convergence issues, Model infeasibility",Less than 1 hour,,1,4,2,3,Relax some constraints,Recommended by peers,No,Maybe,No,User reviews and feedback,"Open Source, Compatible With GAMS, Smaller solving time","I have solved a Resource Allocation Problem for a Communication Network (MINLP). Initially faced problem with ""Bonmin"", Finally solved through GA",,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Non linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear, MINLP",Julia/JuMP and Gurobi Python,"Gurobi, Ipopt","Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","High computational time, Memory limitations, Solver convergence issues, Model infeasibility",1-3 hours,1-30 million,1,,2,,,Best performance for specific models,Yes,Yes,Yes,Benchmark tests,Better support for MINLP,Nonlinear AC Unit Commitment: https://gocompetition.energy.gov/,https://github.com/power-grid-lib and https://gocompetition.energy.gov/challenges/600650/datasets,,"Private Sector (Industry, Business)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)"," PYPSA-BD, PYPSA-TH, Switch-BD, Switch -TH","Gurobi, CPLEX, GLPK","Solver default, Primal or dual simplex","High computational time, Solver convergence issues",1-3 hours,Below 1 million rows / columns,1,,,3,Try to reconfigure the model,Recommended by peers,No,Yes,Maybe,User reviews and feedback,There shall be more features in the academic free versions. ,,,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Power system planning",Linear programming (LP),"IDEEA , IDEEA-WB",CPLEX,Primal or dual simplex,"High computational time, Model infeasibility",Less than 1 hour,,,,,,,Recommended by peers,Yes,Yes,No,Benchmark tests,,,,,"Academic Institution (University, College, Research Institute)"
Multi energy system planning,"Linear programming (LP), Mixed Integer Linear Programming (MILP)",Switch,"Gurobi, CPLEX, Highs, GLPK",Solver default,"High computational time, Model infeasibility",Less than 1 hour,Below 1 million rows / columns,1,2,4,,,Recommended by peers,Yes,Yes,Maybe,"solve a particular problem or run an energy model with different solvers, then based least time needed to solve, a solver is evaluated",,,,,"Academic Institution (University, College, Research Institute)"
Multi energy system planning,Linear programming (LP),energyRt,Gurobi,Solver default,High computational time,Less than 1 hour,1-30 million,,,1,,,Recommended by peers,Yes,Yes,No,Being trained at school / uni with this solver,,,,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Non linear optimal power flow, Power system planning","Linear programming (LP), Non-linear",SWITCH,Gurobi,"Interior Point Method (IPM), or Barrier algorithm","High computational time, Solver convergence issues, Model infeasibility",Less than 1 hour,1-30 million,4,4,2,,,Best performance for specific models,No,Maybe,No,User reviews and feedback,,Multi-node storage problem. It could only be mitigated by reducing the number of nodes.,"Computation time, accuracy of results, ease of usage",,"Academic Institution (University, College, Research Institute)"
"Linear optimal power flow, Multi energy system planning",Mixed Integer Linear Programming (MILP),Developed in-house model,"Highs, CBC",Primal or dual simplex,"High computational time, Solver convergence issues",Less than 1 hour,1-30 million,2,3,1,4,make a different model,"Open source requirement by client, also cost",No,Maybe,Maybe,Benchmark tests,none specific ,"optimisation mid voltage network configuration by setting net openings (switches) to open or closed. We model the complex (non linear) voltage bounds with a linear approximation leading to many binary variables. First used COIN-OR solver, later switched to Gurobi for performance reasons",usually we look at https://plato.asu.edu/bench.html and test run solvers on test problems for the application we are working on to compare results (solution quality) and speed (if relevant),No,"Private Sector (Industry, Business)"
Non linear optimal power flow,Mixed Integer Linear Programming (MILP),PyPSA,Gurobi,Primal or dual simplex,High computational time,1-3 hours,Below 1 million rows / columns,2,3,1,4,,Recommended by peers,Yes,Maybe,Maybe,Being trained at school / uni with this solver,,,,,"Academic Institution (University, College, Research Institute)"
Generation capacity optimisation and expansion,"Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear","EnergyRT, IDEEA","CPLEX, CBC",Solver default,Solver convergence issues,Less than 1 hour,Below 1 million rows / columns,1,2,3,,,Industry standard,Yes,Yes,Maybe,User reviews and feedback,,,,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Unit commitment, Linear optimal power flow, Non linear optimal power flow","Linear programming (LP), Mixed Integer Linear Programming (MILP)",HAMLET,Gurobi,Solver default,High computational time,Less than 1 hour,Below 1 million rows / columns,4,3,2,1,Adjust the problem,Best performance for specific models,No,Yes,Maybe,User reviews and feedback,,,,,"Academic Institution (University, College, Research Institute)"
"Power system planning, Multi energy system planning",Linear programming (LP),PyPSA,Gurobi,"Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","High computational time, Memory limitations",More than 6 hours,30-100 million,4,3,1,2,,Industry standard,Yes,Yes,Maybe,Benchmark tests,easier parallelisation; easier adaptability,,https://plato.asu.edu/bench.html,,"Academic Institution (University, College, Research Institute)"
Multi energy system planning,Mixed Integer Linear Programming (MILP),oemof,Gurobi,Solver default,"High computational time, Memory limitations",1-3 hours,1-30 million,4,1,2,,,Best performance for specific models,Yes,Yes,Yes,User reviews and feedback,"CBC solver does accept parameters, but its functionality is more limited than Gurobi's. Gurobi, on the other hand, offers more robust parameter control. It has a wide range of parameters that can be modified before optimization begins, including time limits, iteration limits, memory limits, and various tolerances. Gurobi's parameters are well-documented and cover areas such as termination conditions, tolerances, cut generation, and distributed algorithms. ",,,,"Academic Institution (University, College, Research Institute)"
"Unit commitment, Linear optimal power flow, Non linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP)",OPERA and own developed models,"Gurobi, CPLEX, GLPK, CBC",Solver default,"High computational time, Memory limitations",3-6 hours,30-100 million,4,3,2,,,Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,,The solver depends on the problem. It also depends of I have the license for a commercial solver. ,Execution times and optimality guarantee ,,Non-Profit Organization
"Economic dispatch, Unit commitment, Non linear optimal power flow, Multi energy system planning, Multi Energy and Materials with Storage",Non-linear,STEVFNs,"SCS, ECOS, GENO","Solver default, Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","High computational time, Memory limitations",Less than 1 hour,Below 1 million rows / columns,2,3,1,4,Try finding an simplification of the model and equations used to describe the assets/ technologies/ processes in the model.,Best performace and free in CVXPY. Comes pre-installed.,Yes,Yes,Maybe,Benchmark tests,"Ability to solve larger problems, especially in time including storage.","I wanted to solve a very large problem with hourly timesteps for a hundred years, with storage. I used the STEVFNs framework to model it in an inherently parallelisable way. Then tried using the new GENO solver that uses GPUs. This was hundreds or thousands of times faster.",The problem defined in my PhD work.,,Government Agency
"Economic dispatch, Unit commitment, Linear optimal power flow, Non linear optimal power flow, Power system planning, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear",Pandapower,"Gurobi, CPLEX, GLPK, CBC","Solver default, Primal or dual simplex, Interior Point Method (IPM), or Barrier algorithm, Running multiple solver instances in parallel","High computational time, Memory limitations, Solver convergence issues, Model infeasibility",3-6 hours,Below 1 million rows / columns,,,,,,Industry standard,,,,Benchmark tests,solver runtime ,It was nilp problem using gams,none,none,Non-Profit Organization
"Economic dispatch, Unit commitment, Linear optimal power flow, Non linear optimal power flow, Multi energy system planning","Linear programming (LP), Mixed Integer Linear Programming (MILP), Non-linear",PowerModels,"Gurobi, Highs, Ipopt","Solver default, Running multiple solver instances in parallel","High computational time, Model infeasibility",1-3 hours,1-30 million,1,1,4,4,"Simplify, fix binaries, decomposition ",Best performance for specific models,Yes,Yes,Maybe,Benchmark tests,Easy call back functionality and parameters control ,Large scale unit commitment with thousands of buses. Test any available solver. ,Time,,"Academic Institution (University, College, Research Institute)"
"Power system planning, Multi energy system planning",Mixed Integer Linear Programming (MILP),PyPSA,"Highs, GLPK",Solver default,"High computational time, Memory limitations",Less than 1 hour,Below 1 million rows / columns,1,3,2,4,,Recommended by peers,Yes,Yes,Maybe,Being trained at school / uni with this solver,,,,,"Academic Institution (University, College, Research Institute)"
"Economic dispatch, Multi energy system planning",Linear programming (LP),PyPSA,Highs,Solver default,"High computational time, Solver convergence issues",Less than 1 hour,Below 1 million rows / columns,4,4,2,1,check problem definition and input parameters,Best performance for specific models,No,Yes,No,Being trained at school / uni with this solver,,,,,Government Agency